Classification of patients with Alzheimer's disease using the arterial pulse spectrum and a multilayer-perceptron analysis

Causes of death in individuals with lifetime major depression: a comprehensive machine learning analysis from a community-based autopsy center

BACKGROUND

Depression can be associated with increased mortality and morbidity, but no studies have investigated the specific causes of death based on autopsy reports. Autopsy studies can yield valuable and detailed information on pathological ailments or underreported conditions. This study aimed to compare autopsy-confirmed causes of death (CoD) between individuals diagnosed with major depressive disorder (MDD) and matched controls. We also analyzed subgroups within our MDD sample, including late-life depression and recurrent depression. We further investigated whether machine learning (ML) algorithms could distinguish MDD and each subgroup from controls based on their CoD.

METHODS

We conducted a comprehensive analysis of CoD in individuals who died from nontraumatic causes. The diagnosis of lifetime MDD was ascertained based on the DSM-5 criteria using information from a structured interview with a knowledgeable informant. Eleven established ML algorithms were used to differentiate MDD individuals from controls by simultaneously analyzing different disease category groups to account for multiple tests. The McNemar test was further used to compare paired nominal data.

RESULTS

The initial dataset included records of 1,102 individuals, among whom 232 (21.1%) had a lifetime diagnosis of MDD. Each MDD individual was strictly paired with a control non-psychiatric counterpart. In the MDD group, the most common CoD were circulatory (67.2%), respiratory (13.4%), digestive (6.0%), and cancer (5.6%). Despite employing a range of ML models, we could not find distinctive CoD patterns that could reliably distinguish individuals with MDD from individuals in the control group (average accuracy: 50.6%; accuracy range: 39-59%). These findings were consistent even when considering factors within the MDD group, such as late-life or recurrent MDD. When comparing groups with paired nominal tests, no differences were found for circulatory (p=0.450), respiratory (p=0.790), digestive (p=1.000), or cancer (p=0.855) CoD.

CONCLUSIONS

Our analysis revealed that autopsy-confirmed CoD exhibited remarkable similarity between individuals with depression and their matched controls, underscoring the existing heterogeneity in the literature. Future research should prioritize more severe manifestations of depression and larger sample sizes, particularly in the context of CoD related to cancer.

Assessing the severity of AstraZeneca COVID-19 vaccine-related side effects through pulse spectrum analysis

AstraZeneca (AZ) vaccine is one of the most common vaccines against COVID-19 used globally. However, adverse reactions post-vaccination have been reported, including severe symptoms and cases of sudden death within several hours. Therefore, this study aimed to establish a database of spectral characteristics of blood pressure waveforms (BPWs) for the AZ vaccine and analyze reactions after vaccine administration using objective physiological signal and symptom analyses for identifying potential differences between heavy and slight groups defined in the study. In total, 24 participants were enrolled in the case-control study. BPW measurements were acquired pre- and post-vaccination. A questionnaire survey on side effects was conducted 5 days after vaccination. The related spectral characteristics of heavy and slight groups were acquired after Fourier transform analysis. Four types of harmonic indexes from BPW signals, including amplitude proportion (Cn), coefficient of variation of Cn (CVn), phase angle (Pn), and standard deviation of Pn (Pn_SD), were derived. The characteristics of harmonic indexes of arterial BPW for the AZ vaccine were in C6 (P = .011), CV2 (P = .027), P5 (P = .009), and P2_SD (P = .027) on the radial pulse. C5 (P = .037), C8 (P = .007), C9 (P = .037), CV5 (P = .015), CV8 (P = .005), and CV9 (P = .028) were significantly different at posttest between heavy and slight groups. In both pretest or posttest, C8 was almost significantly different between slight and heavy groups. More parameters changed significantly post-vaccination, with more severe side effects. Most average values of posttest/pretest of CVn and Pn_SD in the slight group exceeded 100%. All average values of posttest/pretest of CVn and Pn_SD in the heavy group were smaller than 100%. This approach may enable prediction of the risk of reactions post-vaccination to determine suitability of the AZ vaccine and evaluation of side effect severity in vaccinated individuals using pulse analysis to ensure relevant precautions are taken.

Using Arterial Pulse and Laser Doppler Analyses to Discriminate between the Cardiovascular Effects of Different Running Levels

BACKGROUND AND AIMS

Running can induce advantageous cardiovascular effects such as improved arterial stiffness and blood-supply perfusion. However, the differences between the vascular and blood-flow perfusion conditions under different levels of endurance-running performance remains unclear. The present study aimed to assess the vascular and blood-flow perfusion conditions among 3 groups (44 male volunteers) according to the time taken to run 3 km: Level 1, Level 2, and Level 3.

METHODS

The radial blood pressure waveform (BPW), finger photoplethygraphy (PPG), and skin-surface laser-Doppler flowmetry (LDF) signals of the subjects were measured. Frequency-domain analysis was applied to BPW and PPG signals; time- and frequency-domain analyses were applied to LDF signals.

RESULTS

Pulse waveform and LDF indices differed significantly among the three groups. These could be used to evaluate the advantageous cardiovascular effects provided by long-term endurance-running training, such as vessel relaxation (pulse waveform indices), improvement in blood supply perfusion (LDF indices), and changes in cardiovascular regulation activities (pulse and LDF variability indices). Using the relative changes in pulse-effect indices, we achieved almost perfect discrimination between Level 3 and Level 2 (AUC = 0.878). Furthermore, the present pulse waveform analysis could also be used to discriminate between the Level-1 and Level-2 groups.

CONCLUSIONS

The present findings contribute to the development of a noninvasive, easy-to-use, and objective evaluation technique for the cardiovascular benefits of prolonged endurance-running training.

Logistic random forest boosting technique for Alzheimer’s diagnosis

Alzheimer's disease (AD) is a common and well-known neurodegenerative condition that causes cognitive impairment. In the field of medicine, it is the "nervous system" disorder that has received the most attention. Despite this extensive research, there is no treatment or strategy to slow or stop its spread. Nevertheless, there are a variety of options (medication and non-medication alternatives) that may aid in the treatment of AD symptoms at their various phases, thereby enhancing the patient's quality of life. As AD advances over time, it is necessary to treat patients at their various stages appropriately. As a result, detecting and classifying AD phases prior to symptom treatment can be beneficial. Approximately twenty years ago, the rate of progress in the field of machine learning (ML) accelerated dramatically. Using ML methods, this study focuses on early AD identification. The "Alzheimer's Disease Neuroimaging Initiative" (ADNI) dataset was subjected to exhaustive testing for AD identification. The purpose was to classify the dataset into three groups: AD, "Cognitive Normal" (CN), and "Late Mild Cognitive Impairment" (LMCI). In this paper, we present the ensemble model Logistic Random Forest Boosting (LRFB), representing the ensemble of “Logistic Regression” (LR), “Random Forest” (RF), and “Gradient Boost” (GB). The proposed LRFB outperformed LR, RF, GB, “k-Nearest Neighbour” (k-NN), “Multi-Layer Perceptron” (MLP), “Support Vector Machine” (SVM), “AdaBoost” (AB), “Naïve Bayes” (NB), “XGBoost” (XGB), “Decision Tree” (DT), and other ensemble ML models with respect to the performance metrics “Accuracy” (Acc), “Recall” (Rec), “Precision” (Prec), and “F1-Score” (FS).

Discrimination between possible sarcopenia and metabolic syndrome using the arterial pulse spectrum and machine-learning analysis

Sarcopenia is defined as decreased skeletal muscle mass and function, and is an important cause of frailty in the elderly, also being associated with vascular lesions and poor microcirculation. The present study aimed to combine noninvasive pulse measurements, frequency-domain analysis, and machine learning (ML) analysis (1) to determine the effects on the pulse waveform induced by sarcopenia and (2) to develop discriminating models for patients with possible sarcopenia. Radial blood pressure waveform (BPW) signals were measured noninvasively for 1 min in 133 subjects who visited Tri-Service General Hospital for geriatric health checkups. They were assigned to a robust group and a possible-sarcopenia group that combined dynapenia, presarcopenia, and sarcopenia. Two classification methods were used: ML analysis and a self-developed scoring system that used 40 harmonic pulse indices as features: amplitude proportions and their coefficients of variation, and phase angles and their standard deviations. Significant differences were found in several spectral indices of the BPW between possible-sarcopenia and robust subjects. Threefold cross-validation results indicated excellent discrimination performance, with AUC equaling 0.77 when using LDA and 0.83 when using our scoring system. The present noninvasive and easy-to-use measurement and analysis method for detecting sarcopenia-induced changes in the arterial pulse transmission condition could aid the discrimination of possible sarcopenia.

Machine-Learning Classification of Pulse Waveform Quality

Pulse measurements made using wearable devices can aid the monitoring of human physiological condition. Accurate estimation of waveforms is often difficult for nonexperts; motion artifacts may occur during tonometry measurements when the skin-sensor contact pressure is insufficient. An alternative approach is to extract only high-quality pulses for use in index calculations. The present study aimed to determine the effectiveness of using machine-learning analysis in discriminating between high-quality and low-quality pulse waveforms induced by applying different contact pressures. Radial blood pressure waveform (BPW) signals were measured noninvasively in healthy young subjects using a strain-gauge transducer. One-minute-long trains of pulse data were measured when applying the appropriate contact pressure (67.80 ± 1.55 mmHg) and a higher contact pressure (151.80 ± 3.19 mmHg). Eight machine-learning algorithms were employed to evaluate the following 40 harmonic pulse indices: amplitude proportions and their coefficients of variation and phase angles and their standard deviations. Significant differences were noted in BPW indices between applying appropriate and higher skin-surface contact pressures. The present appropriate contact pressure could not only provide a suitable holding force for the wearable device but also helped to maintain the physiological stability of the underlying tissues. Machine-learning analysis provides an effective method for distinguishing between the high-quality and low-quality pulses with excellent discrimination performance (leave-one-subject-out test: random-forest AUC = 0.96). This approach will aid the development of an automatic screening method for waveform quality and thereby improve the noninvasive acquisition reliability. Other possible interfering factors in practical applications can also be systematically studied using a similar procedure.

Multi-Dimensional and Objective Assessment of Motion Sickness Susceptibility Based on Machine Learning

Background

As human transportation, recreation, and production methods change, the impact of motion sickness (MS) on humans is becoming more prominent. The susceptibility of people to MS can be accurately assessed, which will allow ordinary people to choose comfortable transportation and entertainment and prevent people susceptible to MS from entering provocative environments. This is valuable for maintaining public health and the safety of tasks.

Objective

To develop an objective multi-dimensional MS susceptibility assessment model based on physiological indicators that objectively reflect the severity of MS and provide a reference for improving the existing MS susceptibility assessment methods.

Methods

MS was induced in 51 participants using the Coriolis acceleration stimulation. Some portable equipment were used to digitize the typical clinical manifestations of MS and explore the correlations between them and Graybiel's diagnostic criteria. Based on significant objective parameters and selected machine learning (ML) algorithms, several MS susceptibility assessment models were developed, and their performances were compared.

Results

Gastric electrical activity, facial skin color, skin temperature, and nystagmus are related to the severity of MS. Among the ML assessment models based on these variables, the support vector machine classifier had the best performance with an accuracy of 88.24%, sensitivity of 91.43%, and specificity of 81.25%.

Conclusion

The severity of symptoms and signs of MS can be objectively quantified using some indicators. Multi-dimensional and objective assessment models for MS susceptibility based on ML can be successfully established.

Discrimination of the Cognitive Function of Community Subjects Using the Arterial Pulse Spectrum and Machine-Learning Analysis

Early identification of cognitive impairment would allow affected patients to receive care at earlier stage. Changes in the arterial stiffness have been identified as a prominent pathological feature of dementia. This study aimed to verify if applying machine-learning analysis to spectral indices of the arterial pulse waveform can be used to discriminate different cognitive conditions of community subjects. 3-min Radial arterial blood pressure waveform (BPW) signals were measured noninvasively in 123 subjects. Eight machine-learning algorithms were used to evaluate the following 4 pulse indices for 10 harmonics (total 40 BPW spectral indices): amplitude proportion and its coefficient of variation; phase angle and its standard deviation. Significant differences were noted in the spectral pulse indices between Alzheimer’s-disease patients and control subjects. Using them as training data (AUC = 70.32% by threefold cross-validation), a significant correlation (R² = 0.36) was found between the prediction probability of the test data (comprising community subjects at two sites) and the Mini-Mental-State-Examination score. This finding illustrates possible physiological connection between arterial pulse transmission and cognitive function. The present findings from pulse-wave and machine-learning analyses may be useful for discriminating cognitive condition, and hence in the development of a user-friendly, noninvasive, and rapid method for the early screening of dementia.