ECG-based data-driven solutions for diagnosis and prognosis of cardiovascular diseases: A systematic review

Cardiovascular diseases (CVD) are a leading cause of death globally, and result in significant morbidity and reduced quality of life. The electrocardiogram (ECG) plays a crucial role in CVD diagnosis, prognosis, and prevention; however, different challenges still remain, such as an increasing unmet demand for skilled cardiologists capable of accurately interpreting ECG. This leads to higher workload and potential diagnostic inaccuracies. Data-driven approaches, such as machine learning (ML) and deep learning (DL) have emerged to improve existing computer-assisted solutions and enhance physicians' ECG interpretation of the complex mechanisms underlying CVD. However, many ML and DL models used to detect ECG-based CVD suffer from a lack of explainability, bias, as well as ethical, legal, and societal implications (ELSI). Despite the critical importance of these Trustworthy Artificial Intelligence (AI) aspects, there is a lack of comprehensive literature reviews that examine the current trends in ECG-based solutions for CVD diagnosis or prognosis that use ML and DL models and address the Trustworthy AI requirements. This review aims to bridge this knowledge gap by providing a systematic review to undertake a holistic analysis across multiple dimensions of these data-driven models such as type of CVD addressed, dataset characteristics, data input modalities, ML and DL algorithms (with a focus on DL), and aspects of Trustworthy AI like explainability, bias and ethical considerations. Additionally, within the analyzed dimensions, various challenges are identified. To these, we provide concrete recommendations, equipping other researchers with valuable insights to understand the current state of the field comprehensively.

Late Blood Levels of Neurofilament Light Correlate With Outcome in Patients With Traumatic Brain Injury

Neurofilament light (NF-L) is an axonal protein that has shown promise as a traumatic brain injury (TBI) biomarker. Serum NF-L shows a rather slow rise after injury, peaking after 1-2 weeks, although some studies suggest that it may remain elevated for months after TBI. The aim of this study was to examine if plasma NF-L levels several months after the injury correlate with functional outcome in patients who have sustained TBIs of variable initial severity. In this prospective study of 178 patients with TBI and 40 orthopedic injury controls, we measured plasma NF-L levels in blood samples taken at the follow-up appointment on average 9 months after injury. Patients with TBI were divided into two groups (mild [mTBI] vs. moderate-to-severe [mo/sTBI]) according to the severity of injury assessed with the Glasgow Coma Scale upon admission. Recovery and functional outcome were assessed using the Extended Glasgow Outcome Scale (GOSE). Higher levels of NF-L at the follow-up correlated with worse outcome in patients with moderate-to-severe TBI (Spearman's rho = -0.18; p < 0.001). In addition, in computed tomography-positive mTBI group, the levels of NF-L were significantly lower in patients with GOSE 7-8 (median 18.14; interquartile range [IQR] 9.82, 32.15) when compared with patients with GOSE <7 (median 73.87; IQR 32.17, 110.54; p = 0.002). In patients with mTBI, late NF-L levels do not seem to provide clinical benefit for late-stage assessment, but in patients with initially mo/sTBI, persistently elevated NF-L levels are associated with worse outcome after TBI and may reflect ongoing brain injury.

Outlier Analysis for Acute Blood Biomarkers of Moderate and Severe Traumatic Brain Injury

Blood biomarkers have been studied to improve the clinical assessment and prognostication of patients with moderate-severe traumatic brain injury (mo/sTBI). To assess their clinical usability, one needs to know of potential factors that might cause outlier values and affect clinical decision making. In a prospective study, we recruited patients with mo/sTBI (n = 85) and measured the blood levels of eight protein brain pathophysiology biomarkers, including glial fibrillary acidic protein (GFAP), S100 calcium-binding protein B (S100B), neurofilament light (Nf-L), heart-type fatty acid-binding protein (H-FABP), interleukin-10 (IL-10), total tau (T-tau), amyloid β40 (Aβ40) and amyloid β42 (Aβ42), within 24 h of admission. Similar analyses were conducted for controls (n = 40) with an acute orthopedic injury without any head trauma. The patients with TBI were divided into subgroups of normal versus abnormal (n = 9/76) head computed tomography (CT) and favorable (Glasgow Outcome Scale Extended [GOSE] 5-8) versus unfavorable (GOSE <5) (n = 38/42, 5 missing) outcome. Outliers were sought individually from all subgroups from and the whole TBI patient population. Biomarker levels outside Q1 - 1.5 interquartile range (IQR) or Q3 + 1.5 IQR were considered as outliers. The medical records of each outlier patient were reviewed in a team meeting to determine possible reasons for outlier values. A total of 29 patients (34%) combined from all subgroups and 12 patients (30%) among the controls showed outlier values for one or more of the eight biomarkers. Nine patients with TBI and five control patients had outlier values in more than one biomarker (up to 4). All outlier values were > Q3 + 1.5 IQR. A logical explanation was found for almost all cases, except the amyloid proteins. Explanations for outlier values included extremely severe injury, especially for GFAP and S100B. In the case of H-FABP and IL-10, the explanation was extracranial injuries (thoracic injuries for H-FABP and multi-trauma for IL-10), in some cases these also were associated with abnormally high S100B. Timing of sampling and demographic factors such as age and pre-existing neurological conditions (especially for T-tau), explained some of the abnormally high values especially for Nf-L. Similar explanations also emerged in controls, where the outlier values were caused especially by pre-existing neurological diseases. To utilize blood-based biomarkers in clinical assessment of mo/sTBI, very severe or fatal TBIs, various extracranial injuries, timing of sampling, and demographic factors such as age and pre-existing systemic or neurological conditions must be taken into consideration. Very high levels seem to be often associated with poor prognosis and mortality (GFAP and S100B).

Assessing the relevance of mental health factors in fibromyalgia severity: A data-driven case study using explainable AI

BACKGROUND AND OBJECTIVE

Fibromyalgia is a chronic disease that causes pain and affects patients' quality of life. Current treatments focus on pharmacological therapies for pain reduction. However, patients' psychological well-being is also affected, with depression and pain catastrophizing being common. This research addresses the clinicians' need to assess the influence of mental health factors on FM severity compared to pain factors.

METHODS

A co-development study between FM clinicians and data scientists analyzed data from 166 FM-diagnosed patients to assess the influence of mental health factors on FM severity in comparison to pain factors. The study used the Polysymptomatic Distress Scale (PDS) and Fibromyalgia Impact Questionnaire (FIQ) as FM severity indicators and collected 15 variables including regarding demographics, pain intensity perceived, and mental health factors. The team used an author's developed framework to identify the optimal FM severity classifier and explainability by selecting a number of features that lead to obtaining the best classification result. Machine learning classifiers employed in the framework were: decision trees, logistic regression, support vector machines, random forests, AdaBoost, extra trees, and RUSBoost. Explainability analyses were conducted using the following explainable AI techniques: SHapley Additive exPlanations (SHAP), Partial Dependence Plot (PDP), and Mean Decrease Impurity (MDI).

RESULTS

A balanced random forest with 6 features achieved the best performance with PDS (AUC_ROC, mean = 0.81, std = 0.07). Being FIQ the target variable, due to the imbalance in FM severity levels, a binary and a multiclass classification approaches were considered achieving the optimal performance, respectively, a logistic regression classifier (AUC_ROC, mean = 0.83, std = 0.08) with 6 selected features, and a random forest (AUC_ROC, mean = 0.91, std = 0.04) with 8 selected features. Next, the explainability analysis determined mental health factors were found to be more relevant than pain perceived factors for FM severity.

CONCLUSIONS

This study's findings, validated by clinicians, are potentially aligned with FM international guidelines that promote non-pharmacological interventions such as promoting mental well-being of FM patients.

Plasma neurofilament light admission levels and development of axonal pathology in mild traumatic brain injury

BACKGROUND

It is known that blood levels of neurofilament light (NF-L) and diffusion-weighted magnetic resonance imaging (DW-MRI) are both associated with outcome of patients with mild traumatic brain injury (mTBI). Here, we sought to examine the association between admission levels of plasma NF-L and white matter (WM) integrity in post-acute stage DW-MRI in patients with mTBI.

METHODS

Ninety-three patients with mTBI (GCS ≥ 13), blood sample for NF-L within 24 h of admission, and DW-MRI ≥ 90 days post-injury (median = 229) were included. Mean fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) were calculated from the skeletonized WM tracts of the whole brain. Outcome was assessed using the Extended Glasgow Outcome Scale (GOSE) at the time of imaging. Patients were divided into CT-positive and -negative, and complete (GOSE = 8) and incomplete recovery (GOSE < 8) groups.

RESULTS

The levels of NF-L and FA correlated negatively in the whole cohort (p = 0.002), in CT-positive patients (p = 0.016), and in those with incomplete recovery (p = 0.005). The same groups showed a positive correlation with mean MD, AD, and RD (p < 0.001-p = 0.011). In CT-negative patients or in patients with full recovery, significant correlations were not found.

CONCLUSION

In patients with mTBI, the significant correlation between NF-L levels at admission and diffusion tensor imaging (DTI) measurements of diffuse axonal injury (DAI) over more than 3 months suggests that the early levels of plasma NF-L may associate with the presence of DAI at a later phase of TBI.

Trajectories of interleukin 10 and heart fatty acid-binding protein levels in traumatic brain injury patients with or without extracranial injuries

Background

Interleukin 10 (IL-10) and heart fatty acid-binding protein (H-FABP) have gained interest as diagnostic biomarkers of traumatic brain injury (TBI), but factors affecting their blood levels in patients with moderate-to-severe TBI are largely unknown.

Objective

To investigate the trajectories of IL-10 and H-FABP between TBI patients with and without extracranial injuries (ECI); to investigate if there is a correlation between the levels of IL-10 and H-FABP with the levels of inflammation/infection markers C-reactive protein (CRP) and leukocytes; and to investigate if there is a correlation between the admission level of H-FABP with admission levels of cardiac injury markers, troponin (TnT), creatine kinase (CK), and creatine kinase MB isoenzyme mass (CK-MBm).

Materials and methods

The admission levels of IL-10, H-FABP, CRP, and leukocytes were measured within 24 h post-TBI and on days 1, 2, 3, and 7 after TBI. The admission levels of TnT, CK, and CK-MBm were measured within 24 h post-TBI.

Results

There was a significant difference in the concentration of H-FABP between TBI patients with and without ECI on day 0 (48.2 ± 20.5 and 12.4 ± 14.7 ng/ml, p = 0.02, respectively). There was no significant difference in the levels of IL-10 between these groups at any timepoints. There was a statistically significant positive correlation between IL-10 and CRP on days 2 (R = 0.43, p &lt; 0.01) and 7 (R = 0.46, p = 0.03) after injury, and a negative correlation between H-FABP and CRP on day 0 (R = -0.45, p = 0.01). The levels of IL-10 or H-FABP did not correlate with leukocyte counts at any timepoint. The admission levels of H-FABP correlated with CK (R = 0.70, p &lt; 0.001) and CK-MBm (R = 0.61, p &lt; 0.001), but not with TnT.

Conclusion

Inflammatory reactions during the early days after a TBI do not significantly confound the use of IL-10 and H-FABP as TBI biomarkers. Extracranial injuries and cardiac sources may influence the levels of H-FABP in patients with moderate-to-severe TBI.

Transformers for cardiac patient mortality risk prediction from heterogeneous electronic health records

With over 17 million annual deaths, cardiovascular diseases (CVDs) dominate the cause of death statistics. CVDs can deteriorate the quality of life drastically and even cause sudden death, all the while inducing massive healthcare costs. This work studied state-of-the-art deep learning techniques to predict increased risk of death in CVD patients, building on the electronic health records (EHR) of over 23,000 cardiac patients. Taking into account the usefulness of the prediction for chronic disease patients, a prediction period of six months was selected. Two major transformer models that rely on learning bidirectional dependencies in sequential data, BERT and XLNet, were trained and compared. To our knowledge, the presented work is the first to apply XLNet on EHR data to predict mortality. The patient histories were formulated as time series consisting of varying types of clinical events, thus enabling the model to learn increasingly complex temporal dependencies. BERT and XLNet achieved an average area under the receiver operating characteristic curve (AUC) of 75.5% and 76.0%, respectively. XLNet surpassed BERT in recall by 9.8%, suggesting that it captures more positive cases than BERT, which is the main focus of recent research on EHRs and transformers.

Assessing fatigue and sleep in chronic diseases using physiological signals from wearables: A pilot study

Problems with fatigue and sleep are highly prevalent in patients with chronic diseases and often rated among the most disabling symptoms, impairing their activities of daily living and the health-related quality of life (HRQoL). Currently, they are evaluated primarily via Patient Reported Outcomes (PROs), which can suffer from recall biases and have limited sensitivity to temporal variations. Objective measurements from wearable sensors allow to reliably quantify disease state, changes in the HRQoL, and evaluate therapeutic outcomes. This work investigates the feasibility of capturing continuous physiological signals from an electrocardiography-based wearable device for remote monitoring of fatigue and sleep and quantifies the relationship of objective digital measures to self-reported fatigue and sleep disturbances. 136 individuals were followed for a total of 1,297 recording days in a longitudinal multi-site study conducted in free-living settings and registered with the German Clinical Trial Registry (DRKS00021693). Participants comprised healthy individuals (N = 39) and patients with neurodegenerative disorders (NDD, N = 31) and immune mediated inflammatory diseases (IMID, N = 66). Objective physiological measures correlated with fatigue and sleep PROs, while demonstrating reasonable signal quality. Furthermore, analysis of heart rate recovery estimated during activities of daily living showed significant differences between healthy and patient groups. This work underscores the promise and sensitivity of novel digital measures from multimodal sensor time-series to differentiate chronic patients from healthy individuals and monitor their HRQoL. The presented work provides clinicians with realistic insights of continuous at home patient monitoring and its practical value in quantitative assessment of fatigue and sleep, an area of unmet need.

A smart hospital-driven approach to precision pharmacovigilance

Researchers, regulatory agencies, and the pharmaceutical industry are moving towards precision pharmacovigilance as a comprehensive framework for drug safety assessment, at the service of the individual patient, by clustering specific risk groups in different databases. This article explores its implementation by focusing on: (i) designing a new data collection infrastructure, (ii) exploring new computational methods suitable for drug safety data, and (iii) providing a computer-aided framework for distributed clinical decisions with the aim of compiling a personalized information leaflet with specific reference to a drug's risks and adverse drug reactions. These goals can be achieved by using 'smart hospitals' as the principal data sources and by employing methods of precision medicine and medical statistics to supplement current public health decisions.

Digitally Supported Lifestyle Intervention to Prevent Type 2 Diabetes Through Healthy Habits: Secondary Analysis of Long-Term User Engagement Trajectories in a Randomized Controlled Trial

BACKGROUND

Digital health interventions may offer a scalable way to prevent type 2 diabetes (T2D) with minimal burden on health care systems by providing early support for healthy behaviors among adults at increased risk for T2D. However, ensuring continued engagement with digital solutions is a challenge impacting the expected effectiveness.

OBJECTIVE

We aimed to investigate the longitudinal usage patterns of a digital healthy habit formation intervention, BitHabit, and the associations with changes in T2D risk factors.

METHODS

This is a secondary analysis of the StopDia (Stop Diabetes) study, an unblinded parallel 1-year randomized controlled trial evaluating the effectiveness of the BitHabit app alone or together with face-to-face group coaching in comparison with routine care in Finland in 2017-2019 among community-dwelling adults (aged 18 to 74 years) at an increased risk of T2D. We used longitudinal data on usage from 1926 participants randomized to the digital intervention arms. Latent class growth models were applied to identify user engagement trajectories with the app during the study. Predictors for trajectory membership were examined with multinomial logistic regression models. Analysis of covariance was used to investigate the association between trajectories and 12-month changes in T2D risk factors.

RESULTS

More than half (1022/1926, 53.1%) of the participants continued to use the app throughout the 12-month intervention. The following 4 user engagement trajectories were identified: terminated usage (904/1926, 46.9%), weekly usage (731/1926, 38.0%), twice weekly usage (208/1926, 10.8%), and daily usage (83/1926, 4.3%). Active app use during the first month, higher net promoter score after the first 1 to 2 months of use, older age, and better quality of diet at baseline increased the odds of belonging to the continued usage trajectories. Compared with other trajectories, daily usage was associated with a higher increase in diet quality and a more pronounced decrease in BMI and waist circumference at 12 months.

CONCLUSIONS

Distinct long-term usage trajectories of the BitHabit app were identified, and individual predictors for belonging to different trajectory groups were found. These findings highlight the need for being able to identify individuals likely to disengage from interventions early on, and could be used to inform the development of future adaptive interventions.

TRIAL REGISTRATION

ClinicalTrials.gov NCT03156478; https://clinicaltrials.gov/ct2/show/NCT03156478.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

RR2-10.1186/s12889-019-6574-y.

SPECT Image Features for Early Detection of Parkinson's Disease using Machine Learning Methods

Millions of people around the world suffer from Parkinson's disease, a neurodegenerative disorder with no remedy. Currently, the best response to interventions is achieved when the disease is diagnosed at an early stage. Supervised machine learning models are a common approach to assist early diagnosis from clinical data, but their performance is highly dependent on available example data and selected input features. In this study, we explore 23 single photon emission computed tomography (SPECT) image features for the early diagnosis of Parkinson's disease on 646 subjects. We achieve 94 % balanced classification accuracy in independent test data using the full feature space and show that matching accuracy can be achieved with only eight features, including original features introduced in this study. All the presented features can be generated using a routinely available clinical software and are therefore straightforward to extract and apply.

Pharmacogenetics of Bleeding and Thromboembolic Events in Direct Oral Anticoagulant Users

This study aimed to analyze associations between genetic variants and the occurrence of clinical outcomes in dabigatran, apixaban, and rivaroxaban users. This was a retrospective real-world study linking genotype data of three Finnish biobanks with national register data on drug dispensations and healthcare encounters. We investigated several single-nucleotide variants (SNVs) in the ABCG2, ABCB1, CES1, and CYP3A5 genes potentially associated with bleeding or thromboembolic events in direct oral anticoagulant (DOAC) users based on earlier research. We used Cox regression models to compare the incidence of clinical outcomes between carriers and noncarriers of the SNVs or haplotypes. In total, 1,806 patients on apixaban, dabigatran, or rivaroxaban were studied. The ABCB1 c.3435C>T (p.Ile1145=, rs1045642) SNV (hazard ratio (HR) 0.42, 95% confidence interval (CI), 0.18-0.98, P = 0.044) and 1236T-2677T-3435T (rs1128503-rs2032582-rs1045642) haplotype (HR 0.44, 95% CI, 0.20-0.95, P = 0.036) were associated with a reduced risk for thromboembolic outcomes, and the 1236C-2677G-3435C (HR 2.55, 95% CI, 1.03-6.36, P = 0.044) and 1236T-2677G-3435C (HR 5.88, 95% CI, 2.35-14.72, P < 0.001) haplotypes with an increased risk for thromboembolic outcomes in rivaroxaban users. The ABCB1 c.2482-2236G>A (rs4148738) SNV associated with a lower risk for bleeding events (HR 0.37, 95% CI, 0.16-0.89, P = 0.025) in apixaban users. ABCB1 variants are potential factors affecting thromboembolic events in rivaroxaban users and bleeding events in apixaban users. Studies with larger numbers of patients are warranted for comprehensive assessment of the pharmacogenetic associations of DOACs and their relevance for clinical practice.

Gait Disturbances are Associated with Increased Cognitive Impairment and Cerebrospinal Fluid Tau Levels in a Memory Clinic Cohort

Background:

Gait analysis with accelerometers is a relatively inexpensive and easy to use method to potentially support clinical diagnoses of Alzheimer’s disease and other dementias. It is not clear, however, which gait features are most informative and how these measures relate to Alzheimer’s disease pathology.

Objective:

In this study, we tested if calculated features of gait 1) differ between cognitively normal subjects (CN), mild cognitive impairment (MCI) patients, and dementia patients, 2) are correlated with cerebrospinal fluid (CSF) biomarkers related to Alzheimer’s disease, and 3) predict cognitive decline.

Methods:

Gait was measured using tri-axial accelerometers attached to the fifth lumbar vertebra (L5) in 58 CN, 58 MCI, and 26 dementia participants, while performing a walk and dual task. Ten gait features were calculated from the vertical L5 accelerations, following principal component analysis clustered in four domains, namely pace, rhythm, time variability, and length variability. Cognitive decline over time was measured using MMSE, and CSF biomarkers were available in a sub-group.

Results:

Linear mixed models showed that dementia patients had lower pace scores than MCI patients and CN subjects (p < 0.05). In addition, we found associations between the rhythm domain and CSF-tau, especially in the dual task. Gait was not associated with CSF Aβ₄₂ levels and cognitive decline over time as measured with the MMSE.

Conclusion:

These findings suggest that gait — particularly measures related to pace and rhythm — are altered in dementia and have a direct link with measures of neurodegeneration.

Pharmacogenetics of Anticoagulation and Clinical Events in Warfarin-Treated Patients: A Register-Based Cohort Study with Biobank Data and National Health Registries in Finland

Purpose

To assess the association between VKORC1 and CYP2C9 variants and the incidence of adverse drug reactions in warfarin-treated patients in a real-world setting.

Materials and Methods

This was a register-based cohort study (PreMed) linking data from Finnish biobanks, national health registries and patient records between January 1st 2007 and June 30th 2018. The inclusion criteria were: 1) ≥18 years of age, 2) CYP2C9 and VKORC1 genotype information available, 3) a diagnosis of a cardiovascular disease, 4) at least one warfarin purchase, 5) regular INR tests. Eligible individuals were divided into two warfarin sensitivity groups; normal responders, and sensitive and highly sensitive responders based on their VKORC1 and CYP2C9 genotypes. The incidences of clinical events were compared between the groups using Cox regression models.

Results

The cohort consisted of 2508 participants (45% women, mean age of 69 years), of whom 65% were categorized as normal responders and 35% sensitive or highly sensitive responders. Compared to normal responders, sensitive and highly sensitive responders had fewer INR tests below 2 (median: 33.3% vs 43.8%, 95% CI: −13.3%, −10.0%) and more above 3 (median: 18.2% vs 6.7%, 95% Cl: 8.3%, 10.8%). The incidence (per 100 patient-years) of bleeding outcomes was 5.4 for normal responders and 5.6 for the sensitive and highly sensitive responder group (HR=1.03, 95% CI: 0.74, 1.44). The incidence of thromboembolic outcomes was 4.9 and 7.8, respectively (HR=1.48, 95% CI: 1.08, 2.03).

Conclusion

In a real-world setting, genetically sensitive and highly sensitive responders to warfarin had more high INR tests and required a lower daily dose of warfarin than normal responders. However, the risk for bleeding events was not increased in sensitive and highly sensitive responders. Interestingly, the risk of thromboembolic outcomes was lower in normal responders compared to the sensitive and highly sensitive responders.

Trial Registration

NCT04001166.

Admission Levels of Interleukin 10 and Amyloid β 1-40 Improve the Outcome Prediction Performance of the Helsinki Computed Tomography Score in Traumatic Brain Injury

Background: Blood biomarkers may enhance outcome prediction performance of head computed tomography scores in traumatic brain injury (TBI).

Objective: To investigate whether admission levels of eight different protein biomarkers can improve the outcome prediction performance of the Helsinki computed tomography score (HCTS) without clinical covariates in TBI.

Materials and methods: Eighty-two patients with computed tomography positive TBIs were included in this study. Plasma levels of β-amyloid isoforms 1–40 (Aβ40) and 1–42 (Aβ42), glial fibrillary acidic protein, heart fatty acid-binding protein, interleukin 10 (IL-10), neurofilament light, S100 calcium-binding protein B, and total tau were measured within 24 h from admission. The patients were divided into favorable (Glasgow Outcome Scale—Extended 5–8, n = 49) and unfavorable (Glasgow Outcome Scale—Extended 1–4, n = 33) groups. The outcome was assessed 6–12 months after injury. An optimal predictive panel was investigated with the sensitivity set at 90–100%.

Results: The HCTS alone yielded a sensitivity of 97.0% (95% CI: 90.9–100) and specificity of 22.4% (95% CI: 10.2–32.7) and partial area under the curve of the receiver operating characteristic of 2.5% (95% CI: 1.1–4.7), in discriminating patients with favorable and unfavorable outcomes. The threshold to detect a patient with unfavorable outcome was an HCTS > 1. The three best individually performing biomarkers in outcome prediction were Aβ40, Aβ42, and neurofilament light. The optimal panel included IL-10, Aβ40, and the HCTS reaching a partial area under the curve of the receiver operating characteristic of 3.4% (95% CI: 1.7–6.2) with a sensitivity of 90.9% (95% CI: 81.8–100) and specificity of 59.2% (95% CI: 40.8–69.4).

Conclusion: Admission plasma levels of IL-10 and Aβ40 significantly improve the prognostication ability of the HCTS after TBI.

cCOG: A web-based cognitive test tool for detecting neurodegenerative disorders

INTRODUCTION

Web-based cognitive tests have potential for standardized screening in neurodegenerative disorders. We examined accuracy and consistency of cCOG, a computerized cognitive tool, in detecting mild cognitive impairment (MCI) and dementia.

METHODS

Clinical data of 306 cognitively normal, 120 mild cognitive impairment (MCI), and 69 dementia subjects from three European cohorts were analyzed. Global cognitive score was defined from standard neuropsychological tests and compared to the corresponding estimated score from the cCOG tool containing seven subtasks. The consistency of cCOG was assessed comparing measurements administered in clinical settings and in the home environment.

RESULTS

cCOG produced accuracies (receiver operating characteristic-area under the curve [ROC-AUC]) between 0.71 and 0.84 in detecting MCI and 0.86 and 0.94 in detecting dementia when administered at the clinic and at home. The accuracy was comparable to the results of standard neuropsychological tests (AUC 0.69-0.77 MCI/0.91-0.92 dementia).

DISCUSSION

cCOG provides a promising tool for detecting MCI and dementia with potential for a cost-effective approach including home-based cognitive assessments.

Detecting Amyloid Positivity in Elderly With Increased Risk of Cognitive Decline

The importance of early interventions in Alzheimer's disease (AD) emphasizes the need to accurately and efficiently identify at-risk individuals. Although many dementia prediction models have been developed, there are fewer studies focusing on detection of brain pathology. We developed a model for identification of amyloid-PET positivity using data on demographics, vascular factors, cognition, APOE genotype, and structural MRI, including regional brain volumes, cortical thickness and a visual medial temporal lobe atrophy (MTA) rating. We also analyzed the relative importance of different factors when added to the overall model. The model used baseline data from the Finnish Geriatric Intervention Study to Prevent Cognitive Impairment and Disability (FINGER) exploratory PET sub-study. Participants were at risk for dementia, but without dementia or cognitive impairment. Their mean age was 71 years. Participants underwent a brain 3T MRI and PiB-PET imaging. PiB images were visually determined as positive or negative. Cognition was measured using a modified version of the Neuropsychological Test Battery. Body mass index (BMI) and hypertension were used as cardiovascular risk factors in the model. Demographic factors included age, gender and years of education. The model was built using the Disease State Index (DSI) machine learning algorithm. Of the 48 participants, 20 (42%) were rated as Aβ positive. Compared with the Aβ negative group, the Aβ positive group had a higher proportion of APOE ε4 carriers (53 vs. 14%), lower executive functioning, lower brain volumes, and higher visual MTA rating. AUC [95% CI] for the complete model was 0.78 [0.65-0.91]. MRI was the most effective factor, especially brain volumes and visual MTA rating but not cortical thickness. APOE was nearly as effective as MRI in improving detection of amyloid positivity. The model with the best performance (AUC 0.82 [0.71-0.93]) was achieved by combining APOE and MRI. Our findings suggest that combining demographic data, vascular risk factors, cognitive performance, APOE genotype, and brain MRI measures can help identify Aβ positivity. Detecting amyloid positivity could reduce invasive and costly assessments during the screening process in clinical trials.

Admission Levels of Total Tau and β-Amyloid Isoforms 1-40 and 1-42 in Predicting the Outcome of Mild Traumatic Brain Injury

Background: The purpose of this study was to investigate if admission levels of total tau (T-tau) and β-amyloid isoforms 1-40 (Aβ40) and 1-42 (Aβ42) could predict clinical outcome in patients with mild traumatic brain injury (mTBI).

Methods: A total of 105 patients with mTBI [Glasgow Coma Scale (GCS) ≥ 13] recruited in Turku University Hospital, Turku, Finland were included in this study. Blood samples were drawn within 24 h of admission for analysis of plasma T-tau, Aβ40, and Aβ42. Patients were divided into computed tomography (CT)-positive and CT-negative groups. The outcome was assessed 6–12 months after the injury using the Extended Glasgow Outcome Scale (GOSE). Outcomes were defined as complete (GOSE 8) or incomplete (GOSE < 8) recovery. The Rivermead Post Concussion Symptoms Questionnaire (RPCSQ) was also used to assess mTBI-related symptoms. Predictive values of the biomarkers were analyzed independently, in panels and together with clinical parameters.

Results: The admission levels of plasma T-tau, Aβ40, and Aβ42 were not significantly different between patients with complete and incomplete recovery. The levels of T-tau, Aβ40, and Aβ42 could poorly predict complete recovery, with areas under the receiver operating characteristic curve 0.56, 0.52, and 0.54, respectively. For the whole cohort, there was a significant negative correlation between the levels of T-tau and ordinal GOSE score (Spearman ρ = −0.231, p = 0.018). In a multivariate logistic regression model including age, GCS, duration of posttraumatic amnesia, Injury Severity Score (ISS), time from injury to sampling, and CT findings, none of the biomarkers could predict complete recovery independently or together with the other two biomarkers. Plasma levels of T-tau, Aβ40, and Aβ42 did not significantly differ between the outcome groups either within the CT-positive or CT-negative subgroups. Levels of Aβ40 and Aβ42 did not significantly correlate with outcome, but in the CT-positive subgroup, the levels of T-tau significantly correlated with ordinal GOSE score (Spearman ρ = −0.288, p = 0.035). The levels of T-tau, Aβ40, and Aβ42 were not correlated with the RPCSQ scores.

Conclusions: The early levels of T-tau are correlated with the outcome in patients with mTBI, but none of the biomarkers either alone or in any combinations could predict complete recovery in patients with mTBI.

Interleukin 10 and Heart Fatty Acid-Binding Protein as Early Outcome Predictors in Patients With Traumatic Brain Injury

Background: Patients with traumatic brain injury (TBI) exhibit a variable and unpredictable outcome. The proteins interleukin 10 (IL-10) and heart fatty acid-binding protein (H-FABP) have shown predictive values for the presence of intracranial lesions. Aim: To evaluate the individual and combined outcome prediction ability of IL-10 and H-FABP, and to compare them to the more studied proteins S100β, glial fibrillary acidic protein (GFAP), and neurofilament light (NF-L), both with and without clinical predictors. Methods: Blood samples from patients with acute TBI (all severities) were collected <24 h post trauma. The outcome was measured >6 months post injury using the Glasgow Outcome Scale Extended (GOSE) score, dichotomizing patients into: (i) those with favorable (GOSE≥5)/unfavorable outcome (GOSE ≤ 4) and complete (GOSE = 8)/incomplete (GOSE ≤ 7) recovery, and (ii) patients with mild TBI (mTBI) and patients with TBIs of all severities. Results: When sensitivity was set at 95-100%, the proteins' individual specificities remained low. H-FABP showed the best specificity (%) and sensitivity (100%) in predicting complete recovery in patients with mTBI. IL-10 had the best specificity (50%) and sensitivity (96%) in identifying patients with favorable outcome in patients with TBIs of all severities. When individual proteins were combined with clinical parameters, a model including H-FABP, NF-L, and ISS yielded a specificity of 56% and a sensitivity of 96% in predicting complete recovery in patients with mTBI. In predicting favorable outcome, a model consisting IL-10, age, and TBI severity reached a specificity of 80% and a sensitivity of 96% in patients with TBIs of all severities. Conclusion: Combining novel TBI biomarkers H-FABP and IL-10 with GFAP, NF-L and S100β and clinical parameters improves outcome prediction models in TBI.

Impact of a Clinical Decision Support Tool on Dementia Diagnostics in Memory Clinics: The PredictND Validation Study

BACKGROUND

Determining the underlying etiology of dementia can be challenging. Computer- based Clinical Decision Support Systems (CDSS) have the potential to provide an objective comparison of data and assist clinicians.

OBJECTIVES

To assess the diagnostic impact of a CDSS, the PredictND tool, for differential diagnosis of dementia in memory clinics.

METHODS

In this prospective multicenter study, we recruited 779 patients with either subjective cognitive decline (n=252), mild cognitive impairment (n=219) or any type of dementia (n=274) and followed them for minimum 12 months. Based on all available patient baseline data (demographics, neuropsychological tests, cerebrospinal fluid biomarkers, and MRI visual and computed ratings), the PredictND tool provides a comprehensive overview and analysis of the data with a likelihood index for five diagnostic groups; Alzheimer´s disease, vascular dementia, dementia with Lewy bodies, frontotemporal dementia and subjective cognitive decline. At baseline, a clinician defined an etiological diagnosis and confidence in the diagnosis, first without and subsequently with the PredictND tool. The follow-up diagnosis was used as the reference diagnosis.

RESULTS

In total, 747 patients completed the follow-up visits (53% female, 69±10 years). The etiological diagnosis changed in 13% of all cases when using the PredictND tool, but the diagnostic accuracy did not change significantly. Confidence in the diagnosis, measured by a visual analogue scale (VAS, 0-100%) increased (ΔVAS=3.0%, p<0.0001), especially in correctly changed diagnoses (ΔVAS=7.2%, p=0.0011).

CONCLUSION

Adding the PredictND tool to the diagnostic evaluation affected the diagnosis and increased clinicians' confidence in the diagnosis indicating that CDSSs could aid clinicians in the differential diagnosis of dementia.

Metabolic Profiles Help Discriminate Mild Cognitive Impairment from Dementia Stage in Alzheimer's Disease

Accurate differentiation between neurodegenerative diseases is developing quickly and has reached an effective level in disease recognition. However, there has been less focus on effectively distinguishing the prodromal state from later dementia stages due to a lack of suitable biomarkers. We utilized the Disease State Index (DSI) machine learning classifier to see how well quantified metabolomics data compares to clinically used cerebrospinal fluid (CSF) biomarkers of Alzheimer's disease (AD). The metabolic profiles were quantified for 498 serum and CSF samples using proton nuclear magnetic resonance spectroscopy. The patient cohorts in this study were dementia (with a clinical AD diagnosis) (N = 359), mild cognitive impairment (MCI) (N = 96), and control patients with subjective memory complaints (N = 43). DSI classification was conducted for MCI (N = 51) and dementia (N = 214) patients with low CSF amyloid-β levels indicating AD pathology and controls without such amyloid pathology (N = 36). We saw that the conventional CSF markers of AD were better at classifying controls from both dementia and MCI patients. However, quantified metabolic subclasses were more effective in classifying MCI from dementia. Our results show the consistent effectiveness of traditional CSF biomarkers in AD diagnostics. However, these markers are relatively ineffective in differentiating between MCI and the dementia stage, where the quantified metabolomics data provided significant benefit.

"OPTImAL": an ontology for patient adherence modeling in physical activity domain

BACKGROUND

Maintaining physical fitness is a crucial component of the therapeutic process for patients with cardiovascular disease (CVD). Despite the known importance of being physically active, patient adherence to exercise, both in daily life and during cardiac rehabilitation (CR), is low. Patient adherence is frequently composed of numerous determinants associated with different patient aspects (e.g., psychological, clinical, etc.). Understanding the influence of such determinants is a central component of developing personalized interventions to improve or maintain patient adherence. Medical research produced evidence regarding factors affecting patients' adherence to physical activity regimen. However, the heterogeneity of the available data is a significant challenge for knowledge reusability. Ontologies constitute one of the methods applied for efficient knowledge sharing and reuse. In this paper, we are proposing an ontology called OPTImAL, focusing on CVD patient adherence to physical activity and exercise training.

METHODS

OPTImAL was developed following the Ontology Development 101 methodology and refined based on the NeOn framework. First, we defined the ontology specification (i.e., purpose, scope, target users, etc.). Then, we elicited domain knowledge based on the published studies. Further, the model was conceptualized, formalized and implemented, while the developed ontology was validated for its consistency. An independent cardiologist and three CR trainers evaluated the ontology for its appropriateness and usefulness.

RESULTS

We developed a formal model that includes 142 classes, ten object properties, and 371 individuals, that describes the relations of different factors of CVD patient profile to adherence and adherence quality, as well as the associated types and dimensions of physical activity and exercise. 2637 logical axioms were constructed to comprise the overall concepts that the ontology defines. The ontology was successfully validated for its consistency and preliminary evaluated for its appropriateness and usefulness in medical practice.

CONCLUSIONS

OPTImAL describes relations of 320 factors originated from 60 multidimensional aspects (e.g., social, clinical, psychological, etc.) affecting CVD patient adherence to physical activity and exercise. The formal model is evidence-based and can serve as a knowledge tool in the practice of cardiac rehabilitation experts, supporting the process of activity regimen recommendation for better patient adherence.

Correlation of Blood Biomarkers and Biomarker Panels with Traumatic Findings on Computed Tomography after Traumatic Brain Injury

The aim of the study was to examine the ability of eight protein biomarkers and their combinations in discriminating computed tomography (CT)-negative and CT-positive patients with traumatic brain injury (TBI), utilizing highly sensitive immunoassays in a well-characterized cohort. Blood samples were obtained from 160 patients with acute TBI within 24 h of admission. Levels of β-amyloid isoforms 1-40 (Aβ40) and 1-42 (Aβ42), glial fibrillary acidic protein (GFAP), heart fatty-acid binding protein (H-FABP), interleukin 10 (IL-10), neurofilament light (NF-L), S100 calcium-binding protein B (S100B), and tau were measured. Patients were divided into CT-negative (n = 65) and CT-positive (n = 95), and analyses were conducted separately for TBIs of all severities (Glasgow Coma Scale [GCS] score 3-15) and mild TBIs (mTBIs; GCS 13-15). NF-L, GFAP, and tau were the best in discriminating CT-negative and CT-positive patients, both in patients with mTBI and with all severities. In patients with all severities, area under the curve of the receiver operating characteristic (AUC) was 0.822, 0.817, and 0.781 for GFAP, NF-L, and tau, respectively. In patients with mTBI, AUC was 0.720, 0.689, and 0.676, for GFAP, tau, and NF-L, respectively. The best panel of three biomarkers for discriminating CT-negative and CT-positive patients in the group of all severities was a combination of GFAP+H-FABP+IL-10, with a sensitivity of 100% and specificity of 38.5%. In patients with mTBI, the best panel of three biomarkers was H-FABP+S100B+tau, with a sensitivity of 100% and specificity of 46.4%. Panels of biomarkers outperform individual biomarkers in separating CT-negative and CT-positive patients. Panels consisted mainly of different biomarkers than those that performed best as an individual biomarker.

Multivariate Prediction of Hippocampal Atrophy in Alzheimer's Disease

BACKGROUND

Hippocampal atrophy (HA) is one of the biomarkers for Alzheimer's disease (AD).

OBJECTIVE

To identify the best biomarkers and develop models for prediction of HA over 24 months using baseline data.

METHODS

The study included healthy elderly controls, subjects with mild cognitive impairment, and subjects with AD, obtained from the Alzheimer's Disease Neuroimaging Initiative (ADNI 1) and the Australian Imaging Biomarkers and Lifestyle Flagship Study of Ageing (AIBL) databases. Predictor variables included cognitive and neuropsychological tests, amyloid-β, tau, and p-tau from cerebrospinal fluid samples, apolipoprotein E, and features extracted from magnetic resonance images (MRI). Least-mean-squares regression with elastic net regularization and least absolute deviation regression models were tested using cross-validation in ADNI 1. The generalizability of the models including only MRI features was evaluated by training the models with ADNI 1 and testing them with AIBL. The models including the full set of variables were not evaluated with AIBL because not all needed variables were available in it.

RESULTS

The models including the full set of variables performed better than the models including only MRI features (root-mean-square error (RMSE) 1.76-1.82 versus 1.93-2.08). The MRI-only models performed well when applied to the independent validation cohort (RMSE 1.66-1.71). In the prediction of dichotomized HA (fast versus slow), the models achieved a reasonable prediction accuracy (0.79-0.87).

CONCLUSIONS

These models can potentially help identifying subjects predicted to have a faster HA rate. This can help in selection of suitable patients into clinical trials testing disease-modifying drugs for AD.

Impact of a clinical decision support tool on prediction of progression in early-stage dementia: a prospective validation study

Background

In clinical practice, it is often difficult to predict which patients with cognitive complaints or impairment will progress or remain stable. We assessed the impact of using a clinical decision support system, the PredictND tool, to predict progression in patients with subjective cognitive decline (SCD) and mild cognitive impairment (MCI) in memory clinics.

Methods

In this prospective multicenter study, we included 429 patients with SCD (n = 230) and MCI (n = 199) (female 54%, age 67 ± 9, MMSE 28 ± 2) and followed them for at least 12 months. Based on all available patient baseline data (demographics, cognitive tests, cerebrospinal fluid biomarkers, and MRI), the PredictND tool provides a comprehensive overview of the data and a classification defining the likelihood of progression. At baseline, a clinician defined an expected follow-up diagnosis and estimated the level of confidence in their prediction using a visual analogue scale (VAS, 0–100%), first without and subsequently with the PredictND tool. As outcome measure, we defined clinical progression as progression from SCD to MCI or dementia, and from MCI to dementia. Correspondence between the expected and the actual clinical progression at follow-up defined the prognostic accuracy.

Results

After a mean follow-up time of 1.7 ± 0.4 years, 21 (9%) SCD and 63 (32%) MCI had progressed. When using the PredictND tool, the overall prognostic accuracy was unaffected (0.4%, 95%CI − 3.0%; + 3.9%; p = 0.79). However, restricting the analysis to patients with more certain classifications (n = 203), we found an increase of 3% in the accuracy (95%CI − 0.6%; + 6.5%; p = 0.11). Furthermore, for this subgroup, the tool alone showed a statistically significant increase in the prognostic accuracy compared to the evaluation without tool (6.4%, 95%CI 2.1%; 10.7%; p = 0.004). Specifically, the negative predictive value was high. Moreover, confidence in the prediction increased significantly (∆VAS = 4%, p < .0001).

Conclusions

Adding the PredictND tool to the clinical evaluation increased clinicians’ confidence. Furthermore, the results indicate that the tool has the potential to improve prediction of progression for patients with more certain classifications.

Electronic supplementary material

The online version of this article (10.1186/s13195-019-0482-3) contains supplementary material, which is available to authorized users.

Detecting frontotemporal dementia syndromes using MRI biomarkers

Background

Diagnosing frontotemporal dementia may be challenging. New methods for analysis of regional brain atrophy patterns on magnetic resonance imaging (MRI) could add to the diagnostic assessment. Therefore, we aimed to develop automated imaging biomarkers for differentiating frontotemporal dementia subtypes from other diagnostic groups, and from one another.

Methods

In this retrospective multicenter cohort study, we included 1213 patients (age 67 ± 9, 48% females) from two memory clinic cohorts: 116 frontotemporal dementia, 341 Alzheimer's disease, 66 Dementia with Lewy bodies, 40 vascular dementia, 104 other dementias, 229 mild cognitive impairment, and 317 subjective cognitive decline. Three MRI atrophy biomarkers were derived from the normalized volumes of automatically segmented cortical regions: 1) the anterior vs. posterior index, 2) the asymmetry index, and 3) the temporal pole left index. We used the following performance metrics: area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. To account for the low prevalence of frontotemporal dementia we pursued a high specificity of 95%. Cross-validation was used in assessing the performance. The generalizability was assessed in an independent cohort (n = 200).

Results

The anterior vs. posterior index performed with an AUC of 83% for differentiation of frontotemporal dementia from all other diagnostic groups (Sensitivity = 59%, Specificity = 95%, positive likelihood ratio = 11.8, negative likelihood ratio = 0.4). The asymmetry index showed highest performance for separation of primary progressive aphasia and behavioral variant frontotemporal dementia (AUC = 85%, Sensitivity = 79%, Specificity = 92%, positive likelihood ratio = 9.9, negative likelihood ratio = 0.2), whereas the temporal pole left index was specific for detection of semantic variant primary progressive aphasia (AUC = 85%, Sensitivity = 82%, Specificity = 80%, positive likelihood ratio = 4.1, negative likelihood ratio = 0.2). The validation cohort provided corresponding results for the anterior vs. posterior index and temporal pole left index.

Conclusion

This study presents three quantitative MRI biomarkers, which could provide additional information to the diagnostic assessment and assist clinicians in diagnosing frontotemporal dementia.

•

Quantitative MRI biomarkers (API, ASI, and TPL) for detection of FTD and its subtypes.

•

API differentiated FTD from other diagnostic groups with AUC of 83%.

•

ASI and TPL showed highest performance for PPA subtypes.

•

A subcortical bvFTD subtype resembling AD atrophy pattern seems undetectable for MRI.

Prediction models for dementia and neuropathology in the oldest old: the Vantaa 85+ cohort study

Background

We developed multifactorial models for predicting incident dementia and brain pathology in the oldest old using the Vantaa 85+ cohort.

Methods

We included participants without dementia at baseline and at least 2 years of follow-up (N = 245) for dementia prediction or with autopsy data (N = 163) for pathology. A supervised machine learning method was used for model development, considering sociodemographic, cognitive, clinical, vascular, and lifestyle factors, as well as APOE genotype. Neuropathological assessments included β-amyloid, neurofibrillary tangles and neuritic plaques, cerebral amyloid angiopathy (CAA), macro- and microscopic infarcts, α-synuclein pathology, hippocampal sclerosis, and TDP-43.

Results

Prediction model performance was evaluated using AUC for 10 × 10-fold cross-validation. Overall AUCs were 0.73 for dementia, 0.64–0.68 for Alzheimer’s disease (AD)- or amyloid-related pathologies, 0.72 for macroinfarcts, and 0.61 for microinfarcts. Predictors for dementia were different from those in previous reports of younger populations; for example, age, sex, and vascular and lifestyle factors were not predictive. Predictors for dementia versus pathology were also different, because cognition and education predicted dementia but not AD- or amyloid-related pathologies. APOE genotype was most consistently present across all models. APOE alleles had a different impact: ε4 did not predict dementia, but it did predict all AD- or amyloid-related pathologies; ε2 predicted dementia, but it was protective against amyloid and neuropathological AD; and ε3ε3 was protective against dementia, neurofibrillary tangles, and CAA. Very few other factors were predictive of pathology.

Conclusions

Differences between predictors for dementia in younger old versus oldest old populations, as well as for dementia versus pathology, should be considered more carefully in future studies.

Electronic supplementary material

The online version of this article (10.1186/s13195-018-0450-3) contains supplementary material, which is available to authorized users.

Early Levels of Glial Fibrillary Acidic Protein and Neurofilament Light Protein in Predicting the Outcome of Mild Traumatic Brain Injury

The purpose of this study was to correlate the early levels of glial fibrillary acidic protein (GFAP) and neurofilament light protein (NF-L) with outcome in patients with mild traumatic brain injury (mTBI). A total of 107 patients with mTBI (Glasgow Coma Scale ≥13) who had blood samples for GFAP and NF-L available within 24 h of arrival were included. Patients with mTBI were divided into computed tomography (CT)-positive and CT-negative groups. Glasgow Outcome Scale-Extended (GOSE) was used to assess the outcome. Outcomes were defined as complete (GOSE 8) versus incomplete (GOSE <8), and favorable (GOSE 5-8) versus unfavorable (GOSE 1-4). GFAP and NF-L concentrations in blood were measured using ultrasensitive single molecule array technology. Patients with incomplete recovery had significantly higher levels of NF-L compared with those with complete recovery (p = 0.005). The levels of GFAP and NF-L were significantly higher in patients with unfavorable outcome than in patients with favorable outcome (p = 0.002 for GFAP and p < 0.001 for NF-L). For predicting favorable outcome, the area under the receiver operating characteristic curve for GFAP and NF-L was 0.755 and 0.826, respectively. In a multi-variate logistic regression model, the level of NF-L was still a significant predictor for complete recovery (odds ratio [OR] = 1.008; 95% confidence interval [CI], 1.000-1.016). Moreover, the level of NF-L was a significant predictor for complete recovery in CT-positive patients (OR = 1.009; 95% CI, 1.001-1.016). The early levels of GFAP and NF-L are significantly correlated with the outcome in patients with mTBI. The level of NF-L within 24 h from arrival has a significant predictive value in mTBI also in a multi-variate model.

Evaluating combinations of diagnostic tests to discriminate different dementia types

Introduction

We studied, using a data-driven approach, how different combinations of diagnostic tests contribute to the differential diagnosis of dementia.

Methods

In this multicenter study, we included 356 patients with Alzheimer's disease, 87 frontotemporal dementia, 61 dementia with Lewy bodies, 38 vascular dementia, and 302 controls. We used a classifier to assess accuracy for individual performance and combinations of cognitive tests, cerebrospinal fluid biomarkers, and automated magnetic resonance imaging features for pairwise differentiation between dementia types.

Results

Cognitive tests had good performance in separating any type of dementia from controls. Cerebrospinal fluid optimally contributed to identifying Alzheimer's disease, whereas magnetic resonance imaging features aided in separating vascular dementia, dementia with Lewy bodies, and frontotemporal dementia. Combining diagnostic tests increased the accuracy, with balanced accuracies ranging from 78% to 97%.

Discussion

Different diagnostic tests have their distinct roles in differential diagnostics of dementias. Our results indicate that combining different diagnostic tests may increase the accuracy further.

A Decision Support System for Diagnostics and Treatment Planning in Traumatic Brain Injury

Traumatic brain injury (TBI) occurs when an external force causes functional or structural alterations in the brain. Clinical characteristics of TBI vary greatly from patient to patient, and a large amount of data is gathered during various phases of clinical care in these patients. It is hard for clinicians to efficiently integrate and interpret all of these data and plan interventions in a timely manner. This paper describes the technical architecture and functionality of a web-based decision support system (DSS), which not only provides advanced support for visualizing complex TBI data but also predicts a possible outcome by using a state-of-the-art Disease State Index machine-learning algorithm. The DSS is developed by using a three-layered architecture and by employing modern programming principles, software design patterns, and using robust technologies (C#, ASP.NET MVC, HTML5, JavaScript, Entity Framework, etc.). The DSS is comprised of a patient overview module, a disease-state prediction module, and an imaging module. After deploying it on a web-server, the DSS was made available to two hospitals in U.K. and Finland. Afterwards, we conducted a validation study to evaluate its usability in clinical settings. Initial results of the study indicate that especially less experience clinicians may benefit from this type of decision support software tool.

Quantitative EEG Parameters for Prediction of Outcome in Severe Traumatic Brain Injury: Development Study

Monitoring of quantitative EEG (QEEG) parameters in the intensive care unit (ICU) can aid in the treatment of traumatic brain injury (TBI) patients by complementing visual EEG review done by an expert. We performed an explorative study investigating the prognostic value of 59 QEEG parameters in predicting the outcome of patients with severe TBI. Continuous EEG recordings were done on 28 patients with severe TBI in the ICU of Turku University Hospital. We computed a set of QEEG parameters for each patient, and correlated these to patient outcome, measured by dichotomized Glasgow Outcome Scale (GOS) at a follow-up visit between 6 and 12 months, using area under receiver operating characteristic curve (AUC) as a nonlinear correlation measure. For 17 of the 59 QEEG parameters (28.8%), the AUC differed significantly from 0.5, most of these parameters measured EEG power or variability. The best QEEG parameters for outcome prediction were alpha power (AUC = 0.87, P < .01) and variability of the relative fast theta power (AUC = 0.84, P < .01). The results of this study indicate that QEEG parameters provide useful information for predicting outcome in severe TBI. Novel QEEG parameters with potential in outcome prediction were found, the prognostic value of these parameters should be confirmed in later studies. The results also provide further evidence of the usefulness of parameters studied in preexisting studies.

Automatic MRI Quantifying Methods in Behavioral-Variant Frontotemporal Dementia Diagnosis

Aims

We assessed the value of automated MRI quantification methods in the differential diagnosis of behavioral-variant frontotemporal dementia (bvFTD) from Alzheimer disease (AD), Lewy body dementia (LBD), and subjective memory complaints (SMC). We also examined the role of the C9ORF72-related genetic status in the differentiation sensitivity.

Methods

The MRI scans of 50 patients with bvFTD (17 C9ORF72 expansion carriers) were analyzed using 6 quantification methods as follows: voxel-based morphometry (VBM), tensor-based morphometry, volumetry (VOL), manifold learning, grading, and white-matter hyperintensities. Each patient was then individually compared to an independent reference group in order to attain diagnostic suggestions.

Results

Only VBM and VOL showed utility in correctly identifying bvFTD from our set of data. The overall classification sensitivity of bvFTD with VOL + VBM achieved a total sensitivity of 60%. Using VOL + VBM, 32% were misclassified as having LBD. There was a trend of higher values for classification sensitivity of the C9ORF72 expansion carriers than noncarriers.

Conclusion

VOL, VBM, and their combination are effective in differential diagnostics between bvFTD and AD or SMC. However, MRI atrophy profiles for bvFTD and LBD are too similar for a reliable differentiation with the quantification methods tested in this study.

Disease-related determinants are associated with mortality in dementia due to Alzheimer's disease

Background

Survival after dementia diagnosis varies considerably. Previous studies were focused mainly on factors related to demographics and comorbidity rather than on Alzheimer’s disease (AD)-related determinants. We set out to answer the question whether markers with proven diagnostic value also have prognostic value. We aimed to identify disease-related determinants associated with mortality in patients with AD.

Methods

We included 616 patients (50% female; age 67 ± 8 years; mean Mini Mental State Examination score 22 ± 3) with dementia due to AD from the Amsterdam Dementia Cohort. Information on mortality was obtained from the Dutch Municipal Register. We used age- and sex-adjusted Cox proportional hazards analysis to study associations of baseline demographics, comorbidity, neuropsychology, magnetic resonance imaging (MRI) (medial temporal lobe, global cortical and parietal atrophy, and measures of small vessel disease), and cerebrospinal fluid (CSF) (β-amyloid 1–42, total tau, and tau phosphorylated at threonine 181 [p-tau]) with mortality (outcome). In addition, we built a multivariate model using forward selection.

Results

After an average of 4.9 ± 2.0 years, 213 (35%) patients had died. Age- and sex-adjusted Cox models showed that older age (HR 1.29 [95% CI 1.12–1.48]), male sex (HR 1.60 [95% CI 1.22–2.11]), worse scores on cognitive functioning (HR 1.14 [95% CI 1.01-1.30] to 1.31 [95% CI 1.13–1.52]), and more global and hippocampal atrophy on MRI (HR 1.18 [95% CI 1.01-1.37] and HR 1.18 [95% CI 1.02-1.37]) were associated with increased risk of mortality. There were no associations with comorbidity, level of activities of daily living, apolipoprotein E (APOE) ε4 status, or duration of disease. Using forward selection, the multivariate model included a panel of age, sex, cognitive tests, atrophy of the medial temporal lobe, and CSF p-tau.

Conclusions

In this relatively young sample of patients with AD, disease-related determinants were associated with an increased risk of mortality, whereas neither comorbidity nor APOE genotype had any prognostic value.

Electronic supplementary material

The online version of this article (10.1186/s13195-018-0348-0) contains supplementary material, which is available to authorized users.

The Levels of Glial Fibrillary Acidic Protein and Ubiquitin C-Terminal Hydrolase-L1 During the First Week After a Traumatic Brain Injury: Correlations With Clinical and Imaging Findings

BACKGROUND

Glial fibrillary acidic protein (GFAP) and ubiquitin C-terminal hydrolase-L1 (UCH-L1) are promising biomarkers of traumatic brain injury (TBI).

OBJECTIVE

We investigated the relation of the GFAP and UCH-L1 levels to the severity of TBI during the first week after injury.

METHODS

Plasma UCH-L1 and GFAP were measured from 324 consecutive patients with acute TBI and 81 control subject enrolled in a 2-center prospective study. The baseline measures included initial Glasgow Coma Scale (GCS), head computed tomographic (CT) scan at admission, and blood samples for protein biomarkers that were collected at admission and on days 1, 2, 3, and 7 after injury.

RESULTS

Plasma levels of GFAP and UCH-L1 during the first 2 days after the injury strongly correlated with the initial severity of TBI as assessed with GCS. Additionally, levels of UCH-L1 on the seventh day after the injury were significantly related to the admission GCS scores. At admission, both biomarkers were capable of distinguishing mass lesions from diffuse injuries in CT, and the area under the curve of the receiver-operating characteristic curve for prediction of any pathological finding in CT was 0.739 (95% confidence interval, 0.636-0.815) and 0.621 (95% confidence interval, 0.517-0.713) for GFAP and UCH-L1, respectively.

CONCLUSION

These results support the prior findings of the potential role of GFAP and UCH-L1 in acute-phase diagnostics of TBI. The novel finding is that levels of GFAP and UCH-L1 correlated with the initial severity of TBI during the first 2 days after the injury, thus enabling a window for TBI diagnostics with latency.

ABBREVIATIONS

AUC, area under the curveCI, confidence intervalED, emergency departmentGCS, Glasgow Coma ScaleGRAP, glial fibrillary acidic proteinIMPACT, International Mission for Prognosis and Clinical TrialROC, receiver-operating characteristicTBI, traumatic brain injuryTRACK-TBI, Transforming Research and Clinical Knowledge in Traumatic Brain InjuryUCH-L1, ubiquitin C-terminal hydrolase-L1.

Glial Fibrillary Acidic Protein and Ubiquitin C-Terminal Hydrolase-L1 Are Not Specific Biomarkers for Mild CT-Negative Traumatic Brain Injury

Glial fibrillary acidic protein (GFAP) and ubiquitin C-terminal hydrolase-L1 (UCH-L1) have been studied as potential biomarkers of mild traumatic brain injury (mTBI). We report the levels of GFAP and UCH-L1 in patients with acute orthopedic injuries without central nervous system involvement, and relate them to the type of extracranial injury, head magnetic resonance imaging (MRI) findings, and levels of GFAP and UCH-L1 in patients with CT-negative mTBI. Serum UCH-L1 and GFAP were longitudinally measured from 73 patients with acute orthopedic injury on arrival and on days 1, 2, 3, 7 after admission, and on the follow-up visit 3-10 months after the injury. The injury types were recorded, and 71% patients underwent also head MRI. The results were compared with those found in patients with CT-negative mTBI (n = 93). The levels of GFAP were higher in patients with acute orthopedic trauma than in patients with CT-negative mTBI (p = 0.026) on arrival; however, no differences were found on the following days. The levels of UCH-L1 were not significantly different between these two groups at any measured point of time. Levels of GFAP and UCH-L1 were not able to distinguish patients with CT-negative mTBI from patients with orthopedic trauma. Patients with orthopedic trauma and high levels of UCH-L1 or GFAP values may be falsely diagnosed as having a concomitant mTBI, predisposing them to unwarranted diagnostics and unnecessary brain imaging. This casts a significant doubt on the diagnostic value of GFAP and UCH-L1 in cases with mTBI.

Procedures for Evaluating the Adequacy of Anesthesia

Assessing adequacy of anesthesia requires evaluation of its components: hypnosis, analgesia, and neuromuscular transmission. In order to do this, many methods have been developed that process signals representing different modalities. Assessment of hypnosis requires cortical measures of the central nervous system (CNS); methods that assess analgesia concentrate on subcortical and spinal levels of the CNS; and neuromuscular transmission is a peripheral phenomenon. This article presents an overview of the current state of methods available for measuring each of these components. We conclude that, whereas important gains have been made in the area of assessment of hypnosis, mainly owing to the advancement of methods using EEG and auditory evoked potentials, and whereas neuromuscular transmission can be objectively monitored using motor nerve stimulation, assessment of analgesia still contains many challenges.

Prediction of Outcome after Traumatic Brain Injury: Comparison of Disease State Index and IMPACT Calculator

Traumatic brain injury (TBI) is a major cause of death and disability, especially in young adults. A reliable prediction of outcome after TBI is of great importance in clinical practice and research. We aimed to compare performance of the well-established IMPACT calculator and an alternative method, Disease State Index (DSI), in the prediction of six-month outcome after TBI. Performance of the models was evaluated using 2036 patients with moderate or severe TBI from the International Mission for Prognosis and Analysis of Clinical Trials in TBI (IMPACT) database. Prediction performance of both models was similar. The models with more variables provided better performance than the simpler models. This study showed that the DSI is a valid tool with efficient visualizations that can help clinicians with their decision making process in clinical practice.

Graphical Tasks to Measure Upper Limb Function in Patients With Parkinson's Disease: Validity and Response to Dopaminergic Medication

The most widely used method to assess motor functioning in Parkinson's disease (PD) patients is the unified Parkinson's disease rating scale-III (UPDRS-III). The UPDRS-III has limited ability to detect subtle changes in motor symptoms. Alternatively, graphical tasks can be used to provide objective measures of upper limb motor dysfunction. This study investigated the validity of such graphical tasks to assess upper limb function in PD patients and their ability to detect subtle changes in performance. Fourteen PD patients performed graphical tasks before and after taking dopaminergic medication. Graphical tasks included figure tracing, writing, and a modified Fitts' task. The Purdue pegboard test was performed to validate these graphical tasks. Movement time (MT), writing size, and the presence of tremor were assessed. MT on the graphical tasks correlated significantly with performance on the Purdue pegboard test (Spearman's ρ > 0.65; p < 0.05). MT decreased significantly after the intake of dopaminergic medication. Tremor power decreased after taking dopaminergic medication in most PD patients who suffered from tremor. Writing size did not correlate with performance on the Purdue pegboard test, nor did it change after taking medication. Our set of graphical tasks is valid to assess upper limb function in PD patients. MT proved to be the most useful measure for this purpose. The response on dopaminergic medication was optimally reflected by an improved MT on the graphical tasks in combination with a decreased tremor power, whereas writing size did not respond to dopaminergic treatment.

Glial Fibrillary Acidic Protein and Ubiquitin C-Terminal Hydrolase-L1 as Outcome Predictors in Traumatic Brain Injury

OBJECTIVE

Biomarkers ubiquitin C-terminal hydrolase-L1 (UCH-L1) and glial fibrillary acidic protein (GFAP) may help detect brain injury, assess its severity, and improve outcome prediction. This study aimed to evaluate the prognostic value of these biomarkers during the first days after brain injury.

METHODS

Serum UCH-L1 and GFAP were measured in 324 patients with traumatic brain injury (TBI) enrolled in a prospective study. The outcome was assessed using the Glasgow Outcome Scale (GOS) or the extended version, Glasgow Outcome Scale-Extended (GOSE).

RESULTS

Patients with full recovery had lower UCH-L1 concentrations on the second day and patients with favorable outcome had lower UCH-L1 concentrations during the first 2 days compared with patients with incomplete recovery and unfavorable outcome. Patients with full recovery and favorable outcome had significantly lower GFAP concentrations in the first 2 days than patients with incomplete recovery or unfavorable outcome. There was a strong negative correlation between outcome and UCH-L1 in the first 3 days and GFAP levels in the first 2 days. On arrival, both UCH-L1 and GFAP distinguished patients with GOS score 1-3 from patients with GOS score 4-5, but not patients with GOSE score 8 from patients with GOSE score 1-7. For UCH-L1 and GFAP to predict unfavorable outcome (GOS score ≤ 3), the area under the receiver operating characteristic curve was 0.727, and 0.723, respectively. Neither UCHL-1 nor GFAP was independently able to predict the outcome when age, worst Glasgow Coma Scale score, pupil reactivity, Injury Severity Score, and Marshall score were added into the multivariate logistic regression model.

CONCLUSIONS

GFAP and UCH-L1 are significantly associated with outcome, but they do not add predictive power to commonly used prognostic variables in a population of patients with TBI of varying severities.

Detection of sleep-disordered breating with Pressure Bed Sensor

A Pressure Bed Sensor (PBS) can offer an unobtrusive method for sleep monitoring. This study focuses on the detection of the sleep related breathing disorders using a PBS in comparison to the methods used in a sleep laboratory. A newly developed PCA modeling approach for the eight sensor signals of the PBS is evaluated using the Reduced Respiratory Amplitude Index (RRAI) as a central measure. The method computes the respiration amplitude with the Hilbert transform, and then detects the events based on a 20% amplitude reduction from the baseline signal. A similar calculation was used for the sleep laboratory RIP measurements, and both PBS and RIP were compared against the reference based on the nasal flow signal. In the reference RRAI method, the respiratory-disordered events were obtained using RemLogic respiration analyzer to detect over 50% amplitude reduction in the nasal respiratory flow, but removing the RemLogic standard hypopnea event associations on the oxygen desaturation events and the sleep arousals. The movement artifacts were automatically detected based on the movement activity signal of the PBS. Twenty-five (25) out of 28 patients were finally analysed. On average 87% of a night measurement has been covered by the system. The correlation coefficient was 0.92 between the PBS and the reference RRAI, and the performance of the PBS was similar with the RIP belts. Classifying the severity of the sleep related breathing by dividing RRAI in groups according to the severity criteria, the sensitivity was 92% and the specificity was 70% for the PBS. The results suggest that PBS recording can provide an easy and un-obstructive alternative method for the detection of the sleep disordered breathing and thus has a great promise for the home monitoring.

Telemonitoring and Mobile Phone-Based Health Coaching Among Finnish Diabetic and Heart Disease Patients: Randomized Controlled Trial

Background

There is a strong will and need to find alternative models of health care delivery driven by the ever-increasing burden of chronic diseases.

Objective

The purpose of this 1-year trial was to study whether a structured mobile phone-based health coaching program, which was supported by a remote monitoring system, could be used to improve the health-related quality of life (HRQL) and/or the clinical measures of type 2 diabetes and heart disease patients.

Methods

A randomized controlled trial was conducted among type 2 diabetes patients and heart disease patients of the South Karelia Social and Health Care District. Patients were recruited by sending invitations to randomly selected patients using the electronic health records system. Health coaches called patients every 4 to 6 weeks and patients were encouraged to self-monitor their weight, blood pressure, blood glucose (diabetics), and steps (heart disease patients) once per week. The primary outcome was HRQL measured by the Short Form (36) Health Survey (SF-36) and glycosylated hemoglobin (HbA1c) among diabetic patients. The clinical measures assessed were blood pressure, weight, waist circumference, and lipid levels.

Results

A total of 267 heart patients and 250 diabetes patients started in the trial, of which 246 and 225 patients concluded the end-point assessments, respectively. Withdrawal from the study was associated with the patients’ unfamiliarity with mobile phones—of the 41 dropouts, 85% (11/13) of the heart disease patients and 88% (14/16) of the diabetes patients were familiar with mobile phones, whereas the corresponding percentages were 97.1% (231/238) and 98.6% (208/211), respectively, among the rest of the patients (P=.02 and P=.004). Withdrawal was also associated with heart disease patients’ comorbidities—40% (8/20) of the dropouts had at least one comorbidity, whereas the corresponding percentage was 18.9% (47/249) among the rest of the patients (P=.02). The intervention showed no statistically significant benefits over the current practice with regard to health-related quality of life—heart disease patients: beta=0.730 (P=.36) for the physical component score and beta=-0.608 (P=.62) for the mental component score; diabetes patients: beta=0.875 (P=.85) for the physical component score and beta=-0.770 (P=.52) for the mental component score. There was a significant difference in waist circumference in the type 2 diabetes group (beta=-1.711, P=.01). There were no differences in any other outcome variables.

Conclusions

A health coaching program supported with telemonitoring did not improve heart disease patients' or diabetes patients' quality of life or their clinical condition. There were indications that the intervention had a differential effect on heart patients and diabetes patients. Diabetes patients may be more prone to benefit from this kind of intervention. This should not be neglected when developing new ways for self-management of chronic diseases.

Trial Registration

ClinicalTrials.gov NCT01310491; http://clinicaltrials.gov/ct2/show/NCT01310491 (Archived by WebCite at http://www.webcitation.org/6Z8l5FwAM).

Use of home telemonitoring to support multidisciplinary care of heart failure patients in Finland: randomized controlled trial

BACKGROUND

Heart failure (HF) patients suffer from frequent and repeated hospitalizations, causing a substantial economic burden on society. Hospitalizations can be reduced considerably by better compliance with self-care. Home telemonitoring has the potential to boost patients' compliance with self-care, although the results are still contradictory.

OBJECTIVE

A randomized controlled trial was conducted in order to study whether the multidisciplinary care of heart failure patients promoted with telemonitoring leads to decreased HF-related hospitalization.

METHODS

HF patients were eligible whose left ventricular ejection fraction was lower than 35%, NYHA functional class ≥2, and who needed regular follow-up. Patients in the telemonitoring group (n=47) measured their body weight, blood pressure, and pulse and answered symptom-related questions on a weekly basis, reporting their values to the heart failure nurse using a mobile phone app. The heart failure nurse followed the status of patients weekly and if necessary contacted the patient. The primary outcome was the number of HF-related hospital days. Control patients (n=47) received multidisciplinary treatment according to standard practices. Patients' clinical status, use of health care resources, adherence, and user experience from the patients' and the health care professionals' perspective were studied.

RESULTS

Adherence, calculated as a proportion of weekly submitted self-measurements, was close to 90%. No difference was found in the number of HF-related hospital days (incidence rate ratio [IRR]=0.812, P=.351), which was the primary outcome. The intervention group used more health care resources: they paid an increased number of visits to the nurse (IRR=1.73, P<.001), spent more time at the nurse reception (mean difference of 48.7 minutes, P<.001), and there was a greater number of telephone contacts between the nurse and intervention patients (IRR=3.82, P<.001 for nurse-induced contacts and IRR=1.63, P=.049 for patient-induced contacts). There were no statistically significant differences in patients' clinical health status or in their self-care behavior. The technology received excellent feedback from the patient and professional side with a high adherence rate throughout the study.

CONCLUSIONS

Home telemonitoring did not reduce the number of patients' HF-related hospital days and did not improve the patients' clinical condition. Patients in the telemonitoring group contacted the Cardiology Outpatient Clinic more frequently, and on this way increased the use of health care resources.

TRIAL REGISTRATION

Clinicaltrials.gov NCT01759368; http://clinicaltrials.gov/show/NCT01759368 (Archived by WebCite at http://www.webcitation.org/6UFxiCk8Z).

Quantitative evaluation of disease progression in a longitudinal mild cognitive impairment cohort

Several neuropsychological tests and biomarkers of Alzheimer's disease (AD) have been validated and their evolution over time has been explored. In this study, multiple heterogeneous predictors of AD were combined using a supervised learning method called Disease State Index (DSI). The behavior of DSI values over time was examined to study disease progression quantitatively in a mild cognitive impairment (MCI) cohort. The DSI method was applied to longitudinal data from 140 MCI cases that progressed to AD and 149 MCI cases that did not progress to AD during the follow-up. The data included neuropsychological tests, brain volumes from magnetic resonance imaging, cerebrospinal fluid samples, and apolipoprotein E from the Alzheimer's Disease Neuroimaging Initiative database. Linear regression of the longitudinal DSI values (including the DSI value at the point of MCI to AD conversion) was performed for each subject having at least three DSI values available (147 non-converters, 126 converters). Converters had five times higher slopes and almost three times higher intercepts than non-converters. Two subgroups were found in the group of non-converters: one group with stable DSI values over time and another group with clearly increasing DSI values suggesting possible progression to AD in the future. The regression parameters differentiated between the converters and the non-converters with classification accuracy of 76.9% for the slopes and 74.6% for the intercepts. In conclusion, this study demonstrated that quantifying longitudinal patient data using the DSI method provides valid information for follow-up of disease progression and support for decision making.

Standardized handwriting to assess bradykinesia, micrographia and tremor in Parkinson's disease

Objective

To assess whether standardized handwriting can provide quantitative measures to distinguish patients diagnosed with Parkinson's disease from age- and gender-matched healthy control participants.

Design

Exploratory study. Pen tip trajectories were recorded during circle, spiral and line drawing and repeated character ‘elelelel’ and sentence writing, performed by Parkinson patients and healthy control participants. Parkinson patients were tested after overnight withdrawal of anti-Parkinsonian medication.

Setting

University Medical Center Groningen, tertiary care, the Netherlands.

Participants

Patients with Parkinson's disease (n = 10; mean age 69.0 years; 6 male) and healthy controls (n = 10; mean age 68.1 years; 6 male).

Interventions

Not applicable.

Main Outcome Measures

Movement time and velocity to detect bradykinesia and the size of writing to detect micrographia. A rest recording to investigate the presence of a rest-tremor, by frequency analysis.

Results

Mean disease duration in the Parkinson group was 4.4 years and the patients were in modified Hoehn-Yahr stages 1–2.5. In general, Parkinson patients were slower than healthy control participants. Median time per repetition, median velocity and median acceleration of the sentence task and median velocity of the elel task differed significantly between Parkinson patients and healthy control participants (all p<0.0014). Parkinson patients also wrote smaller than healthy control participants and the width of the ‘e’ in the elel task was significantly smaller in Parkinson patients compared to healthy control participants (p<0.0014). A rest-tremor was detected in the three patients who were clinically assessed as having rest-tremor.

Conclusions

This study shows that standardized handwriting can provide objective measures for bradykinesia, tremor and micrographia to distinguish Parkinson patients from healthy control participants.