Can I trust my fake data - A comprehensive quality assessment framework for synthetic tabular data in healthcare

BACKGROUND

Ensuring safe adoption of AI tools in healthcare hinges on access to sufficient data for training, testing and validation. Synthetic data has been suggested in response to privacy concerns and regulatory requirements and can be created by training a generator on real data to produce a dataset with similar statistical properties. Competing metrics with differing taxonomies for quality evaluation have been proposed, resulting in a complex landscape. Optimising quality entails balancing considerations that make the data fit for use, yet relevant dimensions are left out of existing frameworks.

METHOD

We performed a comprehensive literature review on the use of quality evaluation metrics on synthetic data within the scope of synthetic tabular healthcare data using deep generative methods. Based on this and the collective team experiences, we developed a conceptual framework for quality assurance. The applicability was benchmarked against a practical case from the Dutch National Cancer Registry.

CONCLUSION

We present a conceptual framework for quality assuranceof synthetic data for AI applications in healthcare that aligns diverging taxonomies, expands on common quality dimensions to include the dimensions of Fairness and Carbon footprint, and proposes stages necessary to support real-life applications. Building trust in synthetic data by increasing transparency and reducing the safety risk will accelerate the development and uptake of trustworthy AI tools for the benefit of patients.

DISCUSSION

Despite the growing emphasis on algorithmic fairness and carbon footprint, these metrics were scarce in the literature review. The overwhelming focus was on statistical similarity using distance metrics while sequential logic detection was scarce. A consensus-backed framework that includes all relevant quality dimensions can provide assurance for safe and responsible real-life applications of synthetic data. As the choice of appropriate metrics are highly context dependent, further research is needed on validation studies to guide metric choices and support the development of technical standards.

How Real-World Data Can Facilitate the Development of Precision Medicine Treatment in Psychiatry

Precision medicine has the ambition to improve treatment response and clinical outcomes through patient stratification and holds great potential for the treatment of mental disorders. However, several important factors are needed to transform current practice into a precision psychiatry framework. Most important are 1) the generation of accessible large real-world training and test data including genomic data integrated from multiple sources, 2) the development and validation of advanced analytical tools for stratification and prediction, and 3) the development of clinically useful management platforms for patient monitoring that can be integrated into health care systems in real-life settings. This narrative review summarizes strategies for obtaining the key elements-well-powered samples from large biobanks integrated with electronic health records and health registry data using novel artificial intelligence algorithms-to predict outcomes in severe mental disorders and translate these models into clinical management and treatment approaches. Key elements are massive mental health data and novel artificial intelligence algorithms. For the clinical translation of these strategies, we discuss a precision medicine platform for improved management of mental disorders. We use cases to illustrate how precision medicine interventions could be brought into psychiatry to improve the clinical outcomes of mental disorders.

Intelligent digital tools for screening of brain connectivity and dementia risk estimation in people affected by mild cognitive impairment: the AI-Mind clinical study protocol

More than 10 million Europeans show signs of mild cognitive impairment (MCI), a transitional stage between normal brain aging and dementia stage memory disorder. The path MCI takes can be divergent; while some maintain stability or even revert to cognitive norms, alarmingly, up to half of the cases progress to dementia within 5 years. Current diagnostic practice lacks the necessary screening tools to identify those at risk of progression. The European patient experience often involves a long journey from the initial signs of MCI to the eventual diagnosis of dementia. The trajectory is far from ideal. Here, we introduce the AI-Mind project, a pioneering initiative with an innovative approach to early risk assessment through the implementation of advanced artificial intelligence (AI) on multimodal data. The cutting-edge AI-based tools developed in the project aim not only to accelerate the diagnostic process but also to deliver highly accurate predictions regarding an individual's risk of developing dementia when prevention and intervention may still be possible. AI-Mind is a European Research and Innovation Action (RIA H2020-SC1-BHC-06-2020, No. 964220) financed between 2021 and 2026. First, the AI-Mind Connector identifies dysfunctional brain networks based on high-density magneto- and electroencephalography (M/EEG) recordings. Second, the AI-Mind Predictor predicts dementia risk using data from the Connector, enriched with computerized cognitive tests, genetic and protein biomarkers, as well as sociodemographic and clinical variables. AI-Mind is integrated within a network of major European initiatives, including The Virtual Brain, The Virtual Epileptic Patient, and EBRAINS AISBL service for sensitive data, HealthDataCloud, where big patient data are generated for advancing digital and virtual twin technology development. AI-Mind's innovation lies not only in its early prediction of dementia risk, but it also enables a virtual laboratory scenario for hypothesis-driven personalized intervention research. This article introduces the background of the AI-Mind project and its clinical study protocol, setting the stage for future scientific contributions.

Analysis of Lower Extremity Amputations from the SerbVasc Registry

BACKGROUND

Peripheral arterial disease (PAD) and diabetes are the major causes of lower extremity amputations (LEAs) worldwide. Morbidity and mortality in patients with LEAs are high with an associated significant burden on the global health system. The aim of this article is to report the overall morbidity and mortality rates after major and minor LEAs from the Serbian Vascular Registry (SerbVasc), with an analysis of predictive factors that influenced adverse outcomes.

MATERIALS AND METHODS

SerbVasc was created in 2019 as a part of the Vascunet collaboration that is aiming to include all vascular procedures from 21 hospitals in Serbia. Prevalence of diabetes among patients with LEAs, previous revascularization procedures, the degree and the type of foot infection and tissue loss, and overall morbidity and mortality rates were analyzed, with a special reference to mortality predictors.

RESULTS

In the period from January 2020 to December 2022, data on 702 patients with LEAs were extracted from the SerbVasc registry, mean age of 69.06±10.63 years. Major LEAs were performed in 59%, while minor LEAs in 41% of patients. Diabetes was seen in 65.1% of the patients, with 44% of them being on insulin therapy. Before LEA, only 20.3% of patients had previous peripheral revascularization. Soft tissue infection, irreversible acute ischemia, and Fontaine III and IV grade ischemia were the most common causes of above-the-knee amputations while diabetic foot was the most common cause of transphalangeal and toe amputations. The infection rate was 3.7%, the re-amputation rate was 5.7%, and the overall mortality rate was 6.9%, with intrahospital mortality in patients with above-the-knee amputation of 11.1%. The most significant intrahospital mortality predictors were age >65 years (p<0.001), chronic kidney disease (CKD) (p<0.001), ischemic heart disease (IHD) (p=0.001), previous myocardial revascularization (p=0.017), emergency type of admission (p<0.001), not using aspirin (p=0.041), using previous anticoagulation therapy (p=0.003), and postoperative complications (p<0.001).

CONCLUSIONS

The main predictors of increased mortality after LEAs from the SerbVasc registry are age >65 years, CKD, IHD, previous myocardial revascularization, emergency type of admission, not using aspirin, using previous anticoagulation therapy, and postoperative complications. Taking into account high mortality rates after LEAs and a small proportion of previous peripheral revascularization, the work should be done on early diagnosis and timely treatment of PAD hopefully leading to decreased number of LEAs and overall mortality.

CLINICAL IMPACT

Mortality after lower limb amputation from the SerbVasc register is high. A small number of previously revascularized patients is of particular clinical importance, bearing in mind that the main reasons for above-the-knee amputations were irreversible ischemia, Fontaine III and Fontaine IV grade ischemia. Lack of diagnostics procedures and late recognition of patients with PAD, led to subsequent threating limb ischemia and increased amputation rates. The work should be done on early diagnosis and timely treatment of PAD in Serbia, hopefully leading to an increased number of PAD procedures, decreased number of LEAs, and lower overall mortality.

Rare Cause of Leg Edema after Femoropopliteal Bypass Procedure in Patient with Previously Unrecognized Arteriovenous Fistulas: A Case Report

BACKGROUND: Arteriovenous fistulas (AVFs) are pathological, congenital, or acquired communications between the arterial and venous vascular bed. Asymptomatic AVFs do not require surgical or endovascular treatment; however, if they are symptomatic, they must be treated to relieve the patient of symptoms and cardiovascular disorders. CASE PRESENTATION: Our patient had an undiagnosed AVF that became symptomatic after femoropopliteal bypass surgery. We successfully treated these complications with four cover stents. CONCLUSION: The presence of AVFs should be kept in mind in the case of rapidly developing leg edema after revascularization. Endovascular treatment of symptomatic AVF is a safe and effective treatment modality. Treatment of symptomatic AVFs is not only recommended for improving impaired arterial or venous blood flow, but also for preventing recurrent PE.

The age-dependent association of risk factors with pancreatic cancer

BACKGROUND

Pancreatic cancer presents as advanced disease in >80% of patients; yet, appropriate ages to consider prevention and early detection strategies are poorly defined. We investigated age-specific associations and attributable risks of pancreatic cancer for established modifiable and non-modifiable risk factors.

PATIENTS AND METHODS

We included 167 483 participants from two prospective US cohort studies with 1190 incident cases of pancreatic cancer during >30 years of follow-up; 5107 pancreatic cancer cases and 8845 control participants of European ancestry from a completed multicenter genome-wide association study (GWAS); and 248 893 pancreatic cancer cases documented in the US Surveillance, Epidemiology, and End Results (SEER) Program. Across different age categories, we investigated cigarette smoking, obesity, diabetes, height, and non-O blood group in the prospective cohorts; weighted polygenic risk score of 22 previously identified single nucleotide polymorphisms in the GWAS; and male sex and black race in the SEER Program.

RESULTS

In the prospective cohorts, all five risk factors were more strongly associated with pancreatic cancer risk among younger participants, with associations attenuated among those aged >70 years. The hazard ratios comparing participants with three to five risk factors with those with no risk factors were 9.24 [95% confidence interval (CI) 4.11-20.77] among those aged ≤60 years, 3.00 (95% CI 1.85-4.86) among those aged 61-70 years, and 1.46 (95% CI 1.10-1.94) among those aged >70 years (Pheterogeneity = 3×10-5). These factors together were related to 65.6%, 49.7%, and 17.2% of incident pancreatic cancers in these age groups, respectively. In the GWAS and the SEER Program, the associations with the polygenic risk score, male sex, and black race were all stronger among younger individuals (Pheterogeneity ≤0.01).

CONCLUSIONS

Established risk factors are more strongly associated with earlier-onset pancreatic cancer, emphasizing the importance of age at initiation for cancer prevention and control programs targeting this highly lethal malignancy.

Carotid Restenosis Rate After Stenting for Primary Lesions Versus Restenosis After Endarterectomy With Creation of Risk Index

PURPOSE

Carotid artery stenting (CAS) is an option for carotid restenosis (CR) treatment with favorable outcomes. However, CAS has also emerged as an alternative to carotid endarterectomy (CEA) for the management of patients with primary carotid stenosis. This study aimed to report CR rates after CAS was performed in patients with primary lesions versus restenosis after CEA, to identify predictors of CR, and to report both neurological and overall outcomes.

MATERIALS AND METHODS

From January 2000 to September 2018, a total of 782 patients were divided into 2 groups: The CAS (prim) group consisted of 440 patients in whom CAS was performed for primary lesions, and the CAS (res) group consisted of 342 patients with CAS due to restenosis after CEA. Indications for CAS were symptomatic stenosis/restenosis >70% and asymptomatic stenosis/restenosis >85%. A color duplex scan (CDS) of carotid arteries was performed 6 months after CAS, after 1 year, and annually afterward. Follow-up ranged from 12 to 88 months, with a mean follow-up of 34.6±18.0 months.

RESULTS

There were no differences in terms of CR rate between the patients in the CAS (prim) and CAS (res) groups (8.7% vs 7.2%, χ²=0.691, p=0.406). The overall CR rate was 7.9%, whereas significant CR (>70%) rate needing re-intervention was 5.6%, but there was no difference between patients in the CAS (prim) and CAS (res) groups (6.4% vs 4.7%, p=0.351). Six independent predictors for CR were smoking, associated previous myocardial infarction and angina pectoris, plaque morphology, spasm after CAS, the use of FilterWire or Spider Fx cerebral protection devices, and time after stenting. A carotid restenosis risk index (CRRI) was created based on these predictors and ranged from -7 (minimal risk) to +10 (maximum risk); patients with a score >-4 were at increased risk for CR. There were no differences in terms of neurological and overall morbidity and mortality between the 2 groups.

CONCLUSIONS

There was no difference in CR rate after CAS between the patients with primary stenosis and restenosis after CEA. A CRRI score >-4 is a criterion for identifying high-risk patients for post-CAS CR that should be tested in future randomized trials.

Risk Factors and Incidence of Deep Venous Thrombosis in Non-severe Coronavirus Disease-19 Patients

BACKGROUND: The coronavirus disease (COVID-19) is characterized by a high prevalence of deep vein thrombosis (DVT), particularly in its severe form, but the incidence of DVT and risk factors for DVT in non-severe patients are still unknown. METHODS: The study enrolled 118 patients with non-severe COVID-19 infection which did not required hospital admittance. A duplex ultrasound and laboratory test were performed in all the patients after the first negative polymerase chain reaction SARS-CoV-2 test. RESULTS: DVT was identified in 50 (42.4%) patients with a median age of 48 years (interquartile range 30–85 years). Symptomatic DVT was present in 40 (80%) patients and was commonly seen in the Class I calf vein thrombosis (38 patients, 76%) (χ2 = 51.71, p < 0.001). The most significant risk factors for DVT were as follows: Increased C-reactive protein (p = 0.000), fibrinogen (p = 0.000), low lymphocyte count (p = 0.002), obesity (p = 0.017), and neutrophil count (p = 0.042). The multivariable logistic regression analysis revealed that a D-dimer cutoff point of 1253.5 μg/L showed a sensitivity of 92% and a specificity of 71%. CONCLUSION: Patients with increased inflammatory markers and obese patients after a non-severe COVID-19 infection should have an ultrasound examination to ensure early diagnosis of DVT and to prevent the occurrence of any complications.

Severe pacemaker pocket infection during the COVID-19 pandemic, transvenous lead removal

INTRODUCTION

The estimated infection rate after permanent endocardial lead implantation is between 1% and 2%. Pacemaker lead endocarditis is treated with total removal of the infected device and proper antibiotics. In this case report, we present a patient with delayed diagnosis and treatment due to the COVID-19 outbreak.

CASE REPORT

An 88-year-old, pacemaker dependent woman with diagnosed pacemaker pocket infection was admitted to the University Cardiovascular institute. The patient had a prolonged follow-up time due to the COVID-19 outbreak. She missed her routine checkup and came to her local hospital when the generator had already protruded completely, to the point where she held it in her own hand. Transthoracic echocardiogram showed possible vegetations on the lead. Transesophageal echocardiography was not performed due to the COVID-19 pandemic. On the day after the admission the patient underwent transvenous removal of the pacemaker lead using a 9 French gauge rotational extraction sheathe (Cook Medical). The extracted lead was covered in a thin layer of vegetations. Further follow-ups showed good recovery with no complications.

CONCLUSIONS

A case showing delayed treatment of pacemaker pocket infection, due to delayed follow-up time during the COVID-19 pandemic. This patient underwent successful transvenous removal of the infected pacemaker lead, along with adequate antibiotic therapy, which has proven to be the most effective method of treating cardiac device-related endocarditis.

User-Intended Doppler Measurement Type Prediction Combining CNNs With Smart Post-Processing

Spectral Doppler measurements are an important part of the standard echocardiographic examination. These measurements give insight into myocardial motion and blood flow, providing clinicians with parameters for diagnostic decision making. Many of these measurements are performed automatically with high accuracy, increasing the efficiency of the diagnostic pipeline. However, full automation is not yet available because the user must manually select which measurement should be performed on each image. In this work, we develop a pipeline based on convolutional neural networks (CNNs) to automatically classify the measurement type from cardiac Doppler scans. We show how the multi-modal information in each spectral Doppler recording can be combined using a meta parameter post-processing mapping scheme and heatmaps to encode coordinate locations. Additionally, we experiment with several architectures to examine the tradeoff between accuracy, speed, and memory usage for resource-constrained environments. Finally, we propose a confidence metric using the values in the last fully connected layer of the network and show that our confidence metric can prevent many misclassifications. Our algorithm enables a fully automatic pipeline from acquisition to Doppler spectrum measurements. We achieve 96% accuracy on a test set drawn from separate clinical sites, indicating that the proposed method is suitable for clinical adoption.

Analysis of outcomes of single-unit cord blood transplantation with umbilical cord blood units processed with two different red blood cell sedimentation reagents

BACKGROUND

Various processing methodologies are routinely used to reduce volume and red blood cell content of umbilical cord blood (UCB) units collected for hematopoietic stem cell transplantation. There is limited information regarding effects of UCB processing techniques on clinical outcomes.

STUDY DESIGN AND METHODS

Retrospective data analysis compared laboratory and clinical outcomes following single-unit UCB transplantation performed between 1999 and 2015. All UCB units were from St. Louis Cord Blood Bank and all were manually processed with either Hetastarch processed cord blood units (HCB) (n = 661) or PrepaCyte processed cord blood units (PCB) (n = 84). Additional sensitivity analysis focused on units transplanted from 2010 to 2015 and included 105 HCB and 84 PCB.

RESULTS

There were no significant differences in patient characteristics between the two groups. Pre-freeze total nucleated and CD34+ cell counts, cell doses/kg of recipient weight, and total colony-forming units (CFUs) were higher in PCB compared with HCB. Post-thaw, the PCB group had a significantly better total nucleated cell recovery, while there were no significant differences in cell viability, CFU recovery, or CD34+ cell recovery. Primary analysis demonstrated faster neutrophil and platelet engraftment for PCB but no differences in overall survival (OS), whereas sensitivity analysis found no effect of processing method on engraftment, but better OS in the HCB group compared with PCB group.

CONCLUSION

The UCB processing method had no significant impact on engraftment. However, we cannot completely exclude the effect of processing method on OS. Additional studies may be warranted to investigate the potential impact of the PCB processing method on clinical outcomes.

Strain wave pathway to semiconductor-to-metal transition revealed by time-resolved X-ray powder diffraction

One of the main challenges in ultrafast material science is to trigger phase transitions with short pulses of light. Here we show how strain waves, launched by electronic and structural precursor phenomena, determine a coherent macroscopic transformation pathway for the semiconducting-to-metal transition in bistable Ti₃O₅ nanocrystals. Employing femtosecond powder X-ray diffraction, we measure the lattice deformation in the phase transition as a function of time. We monitor the early intra-cell distortion around the light absorbing metal dimer and the long range deformations governed by acoustic waves propagating from the laser-exposed Ti₃O₅ surface. We developed a simplified elastic model demonstrating that picosecond switching in nanocrystals happens concomitantly with the propagating acoustic wavefront, several decades faster than thermal processes governed by heat diffusion.

Method comparison for cardiac image registration of coronary computed tomography angiography and 3-D echocardiography

Treatment decision for coronary artery disease (CAD) is based on both morphological and functional information. Image fusion of coronary computed tomography angiography (CCTA) and three-dimensional echocardiography (3DE) could combine morphology and function into a single image to facilitate diagnosis. Three semiautomatic feature-based methods for CCTA/3DE registration were implemented and applied on CAD patients. Methods were verified and compared using landmarks manually identified by a cardiologist. All methods were found feasible for CCTA/3DE fusion.

Awareness of cord blood collection and the impact on banking

BACKGROUND

Umbilical cord blood (UCB) is an important source of hematopoietic stem cells for transplantation especially in minority populations with limited chances of finding a histocompatible volunteer donor in the registry. UCB has the advantages of early availability, successful outcomes despite some histocompatibility mismatch, and low incidence of chronic graft-versus-host disease. Public cord blood banks that disseminate UCB products for transplant depend on voluntary donation at participating hospitals and obstetrical providers for collection.

PROCEDURE

Using survey questionnaires, we evaluated attitudes toward UCB donation, the frequency of donation, and provider opinions on UCB collection in the greater St. Louis metropolitan area that caters to minority ethnicities in significant numbers.

RESULTS

Our data suggest that nervousness and lack of information regarding the donation and utility of the product were ubiquitous reasons for not donating. Additionally, irrespective of age or level of education, women relied on healthcare providers for information regarding UCB donation. Providers reported primarily time constraints to discussing UCB donation at prenatal visits (54%). Of the interviewees, 62% donated UCB. Fallout due to refusal or preferring private banking was miniscule.

CONCLUSIONS

These results suggest that dedicated personnel focused on disseminating information, obtaining consent, and collecting the UCB product at major hospitals can enrich cord blood banks especially with minority cords. Sustained and focused efforts could improve upon a relatively high wastage rate and ensure a robust supply of UCB products at local public banks.

Abstract AS13: Epidemiologic predictors of pre-treatment CA125 in women with ovarian cancer

Background: CA125 is elevated in 80% of ovarian cancer cases and has proven utility in assessing response to therapy and prognosis. Unfortunately, CA125 is elevated in a variety of benign conditions and only in 50% of early stage ovarian cancers, resulting in a sensitivity and specificity unacceptable for population-based screening. Therefore, understanding factors that influence CA125 at presentation could provide important insights for interpreting CA125 values.

Methods: Using pre-treatment CA125 and detailed epidemiologic data from 12 studies participating in the Ovarian Cancer Association Consortium (OCAC), we evaluated factors previously reported to influence CA125 levels, including age, race, oral contraceptive use, parity, tubal ligation, endometriosis, body mass index (BMI), personal history of breast cancer, or a family history of breast or ovarian cancer. We used linear regression to estimate the association between each variable and CA125 probit scores, which we used to standardize values between studies. Secondary analyses included adjustment for histologic subtype. We also estimated the associations within each study using log-transformed CA125 as the outcome and estimated summary measures using random effects meta-analyses.

Results: Of the 4417 cases included in the analysis, 2918 (66%) were serous, 227 (5%) were mucinous, 484 (11%) were endometrioid, and 258 (6%) were clear cell carcinomas. Median CA125 values varied between studies with a high of 831 U/mL and a low of 271 U/mL. We observed no association between race, oral contraceptive use, tubal ligation, endometriosis, prior breast cancer, and family history of breast cancer and pre-treatment CA125. However, we observed increased pre-treatment CA125 levels with older age (&gt;70 vs. &lt; 50, p=0.03), parity (p=0.01), number of children (p=0.02), obesity (BMI&gt;30 vs. 21-25, p=0.03), and family history of ovarian cancer (p=0.05). Results were similar but attenuated when we calculated summary estimates of the association using meta-analysis of study specific estimates of the association with log-transformed CA125 rather than probit scores. Age and BMI remained predictive of pre-treatment CA125 values after adjustment for histologic subtype. Our data are limited by between site variability in CA125 assays; therefore, validation of these findings is needed with adjustment for type of assay used or in a study population in which all cases had pre-treatment CA125 values measured with the same assay.

Conclusions: Despite these limitations, results from this analysis of over 4000 ovarian cancer cases suggest that age and body size may influence pre-treatment CA125 values.

Citation Format: KL Terry, A Babic, BY Karlan, MT Goodman, D Lambrechts, F Heitz, K Matsuo, I McNeish, T Pejovic, S Kruger Kjaer, PM Webb, E Hogdall, EL Goode, DW Cramer, for the Ovarian Cancer Association Consortium. Epidemiologic predictors of pre-treatment CA125 in women with ovarian cancer [abstract]. In: Proceedings of the 10th Biennial Ovarian Cancer Research Symposium; Sep 8-9, 2014; Seattle, WA. Philadelphia (PA): AACR; Clin Cancer Res 2015;21(16 Suppl):Abstract nr AS13.

Parametric ultrasound and fluoroscopy image fusion for guidance of left ventricle lead placement in cardiac resynchronization therapy

Recent studies show that the response rate to cardiac resynchronization therapy (CRT) could be improved if the left ventricle (LV) is paced at the site of the latest mechanical activation, but away from the myocardial scar. A prototype system for CRT lead placement guidance that combines LV functional information from ultrasound with live x-ray fluoroscopy was developed. Two mean anatomical models, each containing LV epi-, LV endo- and right ventricle endocardial surfaces, were computed from a database of 33 heart failure patients as a substitute for a patient-specific model. The sphericity index was used to divide the observed population into two groups. The distance between the mean and the patient-specific models was determined using a signed distance field metric (reported in mm). The average error values for LV epicardium were [Formula: see text] and for LV endocardium were [Formula: see text]. The validity of using average LV models for a CRT procedure was tested by simulating coronary vein selection in a group of 15 CRT candidates. The probability of selecting the same coronary branch, when basing the selection on the average model compared to a patient-specific model, was estimated to be [Formula: see text]. This was found to be clinically acceptable.

Current practices used for the prevention of central venous catheter-associated infection in hematopoietic stem cell transplantation recipients: a survey from the Infectious Diseases Working Party and Nurses' Group of EBMT

BACKGROUND

Central line-associated bloodstream infection (CLABSI) is one of the most common infectious complications after hematopoietic stem cell transplantation. To prevent this complication, international guidelines recommend the implementation of the CLABSI 'prevention bundle' consisting of hand hygiene, full barrier precautions, cleaning the insertion site with chlorhexidine, avoiding femoral sites for insertion, and removing unnecessary catheters. The aim of this survey was to analyze to what extent European Group for Blood and Marrow Transplantation (EBMT) centers have included the CLABSI prevention bundle in practice.

METHODS

A questionnaire used for data collection was sent to the 545 EBMT centers worldwide, 103 of which responded.

RESULTS

All 5 components of the CLABSI prevention bundle were recorded in 28% of the centers' standard operating procedures (SOP), and 21% of the centers answered that they used all of the bundle components in clinical practice. The most common recommendation absent from the SOP was specification of all the components of full barrier precautions (43% of the centers did not include at least 1 component). Skin disinfection with chlorhexidine before catheter insertion was reported by 66% centers. CLABSI rates were monitored in 21% of centers.

CONCLUSIONS

Although most of the centers lacked 1 or more of the CLABSI prevention bundle components in their SOP, improvements could easily be made by updating the centers' SOP. The first important step is introduction of CLABSI rate monitoring in this high-risk patient population.

Filgrastim XM02 (Tevagrastim®) after autologous stem cell transplantation compared to lenograstim: favourable cost-efficacy analysis

PURPOSE

Granulocyte colony-stimulating factors (G-CSFs), filgrastim and lenograstim, are recognised to be useful in accelerating engraftment after autologous stem cell transplantation. Several forms of biosimilar non-glycosylated G-CSF have been approved by the European Medicines Agency, with limited published data supporting the clinical equivalence in peripheral blood stem cell mobilisation and recovery after autologous stem cell transplantation.

METHOD

With the aim of comparing cost-effective strategies in the use of G-CSF after autologous stem cell transplantation, we retrospectively evaluated 32 patients consecutively treated with biosimilar filgrastim XM02 (Tevagrastim) and 26 with lenograstim. All patients received G-CSF (biosimilar or lenograstim) at a dosage of 5 mcg/kg/day subcutaneously from day 5 to absolute neutrophil count of 1500/mmc for three days.

RESULTS

The median time to absolute neutrophil count engraftment was 11 days for the filgrastim XM02 group and 12 days for the lenograstim group. As for platelets recovery, the median time was 12 days in both groups. The median number of G-CSF vials used for patients was 9.5 for Tevagrastim and 10.5 for lenograstim, reflecting a mean estimated cost of about 556.1 euros for Tevagrastim versus 932.2 euros for lenograstim (p< 0.001). The median days of febrile neutropenia were 1.5 and 1 for filgrastim XM02 and lenograstim, respectively. No adverse event related to the use of XM02 filgrastim was recorded.

CONCLUSION

In our experience, filgrastim XM02 and lenograstim showed comparable efficacy in shortening the period of neutropenia after cytoreduction and autologous stem cell transplantation, with a favourable cost effect for filgrastim XM02.

Donor body mass index is an important factor that affects peripheral blood progenitor cell yield in healthy donors after mobilization with granulocyte-colony-stimulating factor

BACKGROUND

The use of hematopoietic progenitor cell (HPC) transplantation has rapidly expanded in recent years. Currently, several sources of HPCs are available for transplantation including peripheral blood HPCs (PBPCs), cord blood cells, and marrow cells. Of these, PBPC collection has become the major source of HPCs. An important variable in PBPC collection is the response to PBPC mobilization, which varies significantly and sometime causes mobilization failure.

STUDY DESIGN AND METHODS

A retrospective study of 69 healthy donors who underwent PBPC donation by leukapheresis was performed. All of these donors received 10 μg/kg/day or more granulocyte-colony-stimulating factor (G-CSF) for 5 days before PBPC harvest. Donor factors were evaluated and correlated with mobilization responses, as indicated by the precollection CD34 count (pre-CD34).

RESULTS

Donors with a pre-CD34 of more than 100 × 10(6) /L had higher body mass index (BMI) compared with donors whose pre-CD34 was 38 × 10(6) to 99 × 10(6) /L or less than 38 × 10(6) /L (32.0 ± 1.04 kg/m(2) vs. 28.7 ± 0.93 kg/m(2) vs. 25.9 ± 1.27 kg/m(2) , respectively; p < 0.05). In addition, donors with high BMIs had higher pre-CD34 on a per-kilogram-of-body-weight basis compared with donors with low BMIs.

CONCLUSION

BMI is an important factor that affects donor's response to mobilization and consequently the HPC yield. This effect may be due to a relatively high dose of G-CSF administered to donors with higher BMI or due to the presence of unknown intrinsic factors affecting mobilization that correlate with the amount of adipose tissue in each donor.

Synergy between medical informatics and bioinformatics: facilitating genomic medicine for future health care

In this paper, we review the results of BIOINFOMED, a study funded by the European Commission (EC) with the purpose to analyse the different issues and challenges in the area where Medical Informatics and Bioinformatics meet. Traditionally, Medical Informatics has been focused on the intersection between computer science and clinical medicine, whereas Bioinformatics have been predominantly centered on the intersection between computer science and biological research. Although researchers from both areas have occasionally collaborated, their training, objectives and interests have been quite different. The results of the Human Genome and related projects have attracted the interest of many professionals, and introduced new challenges that will transform biomedical research and health care. A characteristic of the 'post genomic' era will be to correlate essential genotypic information with expressed phenotypic information. In this context, Biomedical Informatics (BMI) has emerged to describe the technology that brings both disciplines (BI and MI) together to support genomic medicine. In recognition of the dynamic nature of BMI, institutions such as the EC have launched several initiatives in support of a research agenda, including the BIOINFOMED study.

Cytological grading of breast cancer in Giemsa-stained fine needle aspiration smears

A potential cytological nuclear grading based on a semi-quantitative evaluation of three basic nuclear features, size of cell nuclei, anisonucleosis and the proportion of nucleoli-containing-nuclei, was tested on 74 Giemsa-stained fine needle aspiration of breast smears for its reliability in establishing the malignant potential of breast cancer. The prognostic impact of DNA-ploidy and S-phase fraction was also assessed. A good correlation between the three basic nuclear features, DNA-ploidy, S-phase fraction, cytological nuclear grade and histological grade, was shown. Using the cytological nuclear grade proposed, correct classification of cases between low histological grade (HG I) and high histological grade (HG II + HG III) was achieved in 79.73%. A statistically significant difference in 5-year survival rate was also observed between low malignancy grade and high malignancy grade breast cancer patients, regardless of the grading method used. DNA-ploidy and S-phase fraction were not statistically significant in establishing the malignant potential of breast cancer.

Quantitative collagen as a golden standard in differential diagnosing of fibrotic changes in liver tissue

Determining a presence and degree of liver fibrosis provides means for diagnosing disease related processes. We have used two data mining methods, discriminant and regression analyses, to acquire knowledge from the data of 211 patients. We have shown and discussed that quantitative collagen has a distinguished discriminating power and can serve as a golden standard. We have additionally succeeded to obtain a formula consisting of standardised blood tests that can replace quantitative collagen. Practical implications of this is a non-invasive and cost efficient patient examination. All the results are now left for clinical evaluation and so is the current way of histopathological classifications.

Take care: guidelines for patients with chronic hepatitis C

Alcohol consumption has significant impact on the condition of the liver, by itself, and even more in conjunction with other liver diseases such as chronic hepatitis C. Drinking habits might be delicate issues to address and could harm otherwise satisfying communication. Therefore, we intended to outline guidelines for advising hepatitis C patients concerning alcohol consumption. Analysis of a relatively limited knowledge base revealed the complexity of the disease rather than statistically significant findings regarding consumption. Thus, we instead chose to suggest a set of patient educational guidelines, which could be implemented on the Internet, hypothesizing that a better informed patient will be more able to comply with restrictions concerning alcohol consumption. A brief ad hoc evaluation pointed out Internet as a favourable media to present the information. We also suggest a tentative algorithm for further development of clinical decision support systems addressing monitoring of chronic hepatitis C patients.

Knowledge discovery for advanced clinical data management and analysis

Knowledge discovery is a broad research field in which methods are developed to support discovery of novel and potentially useful knowledge from clinical databases and registers in systems for patient care. However, the techniques available are not readily applicable in medical domains, due to, among other reasons, low user friendliness and lack of proper methodological background. Data mining approaches to be explored and improved are predictive modelling, segmentation, dependency modelling, summarization, and change and deviation detection/modelling (in data or knowledge). Another and original contribution of the research is to build up efficient feedback loops. Human experts and available domain expert systems could provide suggestions as how to improve all major steps in the knowledge discovery process such as evaluation of knowledge, choice of data mining methods and data input. A long tradition of collecting and maintaining clinical and administrative data could be found in fields of oncology, cardiology, coronary surgery, social and primary health care medicine. All these areas, that gather data over long periods of time, could benefit from knowledge discovery.

Quantitative data analysis for exploring outcomes in cardiac surgery

The article focuses on possibilities of statistical knowledge exploration to predict outcomes of surgical treatments. The outcomes were defined in relation to the measured peri- and intraoperative data, as well as follow-up patient questionnaire. Clinical consequences are expected in terms of a smaller data set with a better ability to predict the surgery outcomes and a better cost performance. The important questions that could discriminate quality of life (QoL) were: Relief from surgery?, Has cardiac surgery effected earlier symptoms? Work capacity? Consultations after the surgery? The performed data analysis proved to be efficient in the complex data set that was collected. Pain relief was identified to be significant, while relations between measured blood laboratory profile and later QoL were weak.

Knowledge representation forms for data mining methodologies as applied in thoracic surgery

Typical ways of disseminating and using results of clinical research are scientific journals and reports. Presentation forms are condensed and comprehensible mainly to the experts following the specific topics. A vast amount of information remains unutilized due to the complex form of presenting the knowledge. Subject of this research is to explore possibilities of representation and also visualization of the results obtained using data mining methodologies. The intention is to formulate more than scientific ways to communicate facts that are of interest for the clinicians, medical students and even patients. Internet technologies as already widely established media support knowledge representation forms such as hypertext documents and structured knowledge components. The "Assist Me" decision support system for surgical treatment of cardiac patients integrates several forms of data mining and representation methodologies. We are showing a feasibility study in which scientific outcomes were forwarded to a broad group of potential users.

Liver guide for monitoring of chronic hepatitis C

The severity of chronic hepatitis C infection in the individual patient is monitored using blood laboratory findings and liver biopsy. If blood test results could be shown to provide sufficient information concerning the disease, the invasive procedure of liver biopsy could perhaps be avoided in some instances. This study assessed the clinical relevance of blood laboratory tests for detecting disease-related changes in the liver. Histopathological classification was used to assign class membership of the patients and data mining operations were performed in an elaborate way on 19 different data sets. Disease activity could be detected by a small set of blood tests. Extended sets could identify more severe changes, but failed to distinguish them. The extracted rules are implemented as a part of the knowledge base of a corresponding decision support system aimed at specialists and general practitioners.

Aspects of certainty in patient classification using a health-related quality-of-life instrument in inflammatory bowel disease

The study has focused on deriving a certainty measure for the classification of disease activity in patients suffering from inflammatory bowel disease (IBD). The aim is to build an Internet-based health-related quality-of-life (HRQoL) questionnaire to continuously monitor a patient's condition. Data from 109 patients was collected four times in intervals of three months, using a standardized disease-specific quality-of-life questionnaire, the Rating Form of IBD Patient Concern (RFIPC), extended with 11 additional questions. Correlation analysis showed that the RFIPC items along with "general wellbeing" were highly correlated (significance < 0.001). Factor analysis confirmed this high correlation and only one factor was identified among those variables. Multivariate discriminant analysis was successful to 78.1% in classifying between cases of remission and relapse. Implementation of a smooth threshold function decreased the classification error. However, discrimination regarding change in disease activity over time has to be further improved.

Assessing an AI knowledge-base for asymptomatic liver diseases

Discovering not yet seen knowledge from clinical data is of importance in the field of asymptomatic liver diseases. Avoidance of liver biopsy which is used as the ultimate confirmation of diagnosis by making the decision based on relevant laboratory findings only, would be considered an essential support. The system based on Quinlan's ID3 algorithm was simple and efficient in extracting the sought knowledge. Basic principles of applying the AI systems are therefore described and complemented with medical evaluation. Some of the diagnostic rules were found to be useful as decision algorithms i.e. they could be directly applied in clinical work and made a part of the knowledge-base of the Liver Guide, an automated decision support system.

Diagnostic systems for assessing alveolar bone loss

Radiological diagnostics serves as a basic monitoring technique for alveolar bone loss which is a severe consequence of periodontal disease. To evaluate efficacy of Conventional Visual Radiography (CVR), and to assess a complete clinical status, we had used two more diagnostic systems. These are Digital Subtraction Radiography (DSR) and Probing Pocket Depth (PPD). Experimental Periodontitis was studied in 20 beagle dogs based on the measurements taken in the beginning (baseline), and before (11th month) and after the medical treatment (12th month). Data analyses pointed out the same clinical trend, i.e. a significant bone loss prior to medical treatment and its recovery to the initial state. Differences in metrics and measurement errors could be identified as causes for discrepancies between the systems, but a relationship between the CVR and PPD is worth of further research, as these systems do not appear to be entirely compatible, but rather complementary to each other.

Machine learning for the quality of life in inflammatory bowel disease

Presence of a chronic disease influences patients' lives and reinforces demands to accept and then cope with the illness. In the case of inflammatory bowel disease, quality of life greatly differs through phases of remissions and relapses. Could the quality of life questionnaire tell the difference? In this study we are disclosing possibilities of assessing patients' perspectives by analysing analogue scale statements regarding concerns and worries related to ulcerative colitis. Some two hundred Swedish patients, 3/4 in remission and 1/4 in relapse, filled out a booklet containing 36 statements. To characterise the disease activity, we have used multivariate discrimination. To structure and describe in details paths distinguishing the remission from relapse, we have used an artificial intelligence procedure. Applications of the CART (Classification And Regression Trees) algorithm resulted in a set of classifiers which are, based on the similar subsets of significant variables, i.e. statements. Best reached classification accuracy did not exceed 80% in any case. Other classifiers namely, K-nearest-neighbour (KNN), Learning Vector Quantization (LVQ) and Back Propagation Neural Network (BPNN) confirmed that outcome. An expectation that the disease activity should clearly speak throughout the questionnaire held for a certain number of the observations such as pain and suffering, loss of bowel control, dying early, feeling alone, ability to have children, being treated as different and concerns regarding the medication. To highlight the difference of incorrect 20%, K-means clustering was performed. The results settled a basis for a hypothesis that the studied quality of life instrument captures more than the disease activity.

Hemopump treatment in patients with postcardiotomy heart failure

BACKGROUND

This study examined the use of the Hemopump to treat low cardiac output syndrome after cardiopulmonary bypass.

METHODS

We used the Hemopump temporary cardiac assist system in 29 patients with severe left ventricular dysfunction after open heart operations from September 1991 to November 1994.

RESULTS

Five patients were excluded from the study due to initial patient/device-related problems. Ten patients died in the operating room or early during the stay in the intensive care unit due to progressive biventricular failure. Fourteen patients (58.3%) were weaned from the device, and all of them were later discharged. In a subgroup of patients (54%) in whom we had a more aggressive approach for early insertion of the pump, the survival rate was 85%. Preoperative Higging risk score was significantly related to survival.

CONCLUSIONS

The Hemopump can effectively unload a failing left ventricle with preservation of multiorgan perfusion. A minor decrease in kidney function was observed in most patients, but none of the surviving patients needed hemodialysis. One patient required a short period of peritoneal dialysis to get rid of fluid overload. Hemolysis or platelet dysfunction was not a clinical problem.

Machine learning to support diagnostics in the domain of asymptomatic liver disease

Machine learning procedures, in unsupervised and supervised manner, can enable their users to achieve knowledge hardly comprehensible by even the best experts. This is true also if the clinical knowledge has been carefully assembled in a prospective way. A data set including 165 patients with elevated routine laboratory tests was extensively studied according to clinical history, laboratory profile and liver biopsy. Unsupervised learning by Kohonen feature map disclosed 4 groups of patients: the largest one with no or slight histopathological changes (116) and three smaller, more homogenous, with more diseased patients. Standardized histopathological scorings of the liver specimens defined patients into two groups. Fifty-eight of them were, according to the analysis, recommended for a liver biopsy, due to more severe degrees of inflammation and fibrosis. One-hundred and seven of the patients, in whom liver biopsy was retrospectively considered unnecessary, had only minor degrees of inflammation, fibrosis and/or steatosis. Supervised learning, using the inductive systems based on Quinlan's ID3 and CART algorithms, extracted knowledge in the form of decision trees. This approach could define a need for biopsy either with a very few significant findings or by pathways containing quotients and multiplications of the different basic items. These procedures were analyzed and compared for their theoretical and applicative performances. The cluster and Fischerian discriminant analyses were performed in order to compare the classification performance. The medical appropriateness of the obtained results is satisfying, therefore decision support systems, outlined in this study, should be evaluated in wider clinical practice. To achieve this goal, an example of a Medical Logical Module (MLM), based on the Arden Syntax, is given.

A statistically rule-based decision support system for the management of patients with suspected liver disease

The paper describes how a decision support system in liver diseases, mostly oriented to prediction of the necessity for liver biopsy, has been developed. The system designed is a hybrid one and consists of two parts: logical and statistical. The logical part contains rules, formulated on the basis of current medical knowledge, which enables recognition of clear cases; diseased or non-diseased. The unclear cases are classified on the basis of rules statistically extracted from databases. These rules have been reached after a comprehensive exploratory analysis of the sample of 165 patients with slightly to moderately raised levels of routine liver tests but without signs or symptoms of liver diseases. The extracted decision diagrams which simulate traditional medical diagnosis conduct have been found to be superior to discriminant analysis and probabilistic inductive learning. They use only a limited number of laboratory tests to detect the necessity for biopsy.

Methods for knowledge extraction from a clinical database on liver diseases

We performed exploratory data analysis (EDA) to examine the hidden structure in liver disease data. The purpose was to demonstrate the potential of statistical techniques for extracting knowledge from an active HIS (hospital information system) database with decision support. The goal is to give strong support to the creation of new rules or "tuning" of old rules in the knowledge base. This would facilitate utilization of large patient databases, now commonly available, to help build/update decision support systems for improved patient care. Several statistical techniques were investigated. Stepwise discriminant analysis was found to be a good method in discriminating among different disease classes. Results showed that classification strength of a few (3) variables was similar to all the available (19) variables. Other important issues in the work are treatment of missing values as well as atypical values in medical databases. In estimating missing values we utilized both statistical methods and artificial intelligence approaches. Both these approaches were promising in the estimation of missing values. The study showed that several statistical approaches are possible for knowledge extraction from clinical data collected retrospectively.

Integrated approach for designing medical decision support systems with knowledge extracted from clinical databases by statistical methods

In clinical research data is often studied by a particular method without previous analysis of quality or semantic contents which could link clinical database and data analytical (e.g. statistical) procedures. In order to avoid bias caused by this situation, we propose that the analysis of medical data should be divided into two main steps. In the first one we concentrate on conducting the quality, semantic and structure analyses. In the second step our aim is to build an appropriate dictionary of data analysis methods for further knowledge extraction. Methods like robust statistical techniques, procedures for mixed continuous and discrete data, fuzzy linguistic approach, machine learning and neural networks can be included. The results may be evaluated both using test samples and applying other relevant data-analytical techniques to the particular problem under the study.