Evaluation of deep learning-based depression detection using medical claims data.

Human accuracy in diagnosing psychiatric disorders is still low. Even though digitizing health care leads to more and more data, the successful adoption of AI-based digital decision support (DDSS) is rare. One reason is that AI algorithms are often not evaluated based on large, real-world data. This research shows the potential of using deep learning on the medical claims data of 812,853 people between 2018 and 2022, with 26,973,943 ICD-10-coded diseases, to predict depression (F32 and F33 ICD-10 codes). The dataset used represents almost the entire adult population of Estonia. Based on these data, to show the critical importance of the underlying temporal properties of the data for the detection of depression, we evaluate the performance of non-sequential models (LR, FNN), sequential models (LSTM, CNN-LSTM) and the sequential model with a decay factor (GRU-Δt, GRU-decay). Furthermore, since explainability is necessary for the medical domain, we combine a self-attention model with the GRU decay and evaluate its performance. We named this combination Att-GRU-decay. After extensive empirical experimentation, our model (Att-GRU-decay), with an AUC score of 0.990, an AUPRC score of 0.974, a specificity of 0.999 and a sensitivity of 0.944, proved to be the most accurate. The results of our novel Att-GRU-decay model outperform the current state of the art, demonstrating the potential usefulness of deep learning algorithms for DDSS development. We further expand this by describing a possible application scenario of the proposed algorithm for depression screening in a general practitioner (GP) setting-not only to decrease healthcare costs, but also to improve the quality of care and ultimately decrease people's suffering.

The NORDeHEALTH 2022 Patient Survey: Cross-Sectional Study of National Patient Portal Users in Norway, Sweden, Finland, and Estonia.

BACKGROUND: Although many surveys have been conducted on patients accessing their own health records in recent years, there is a limited amount of nationwide cross-country data available on patients' views and preferences. To address this gap, an international survey of patient users was conducted in the Nordic eHealth project, NORDeHEALTH. OBJECTIVE: We aimed to investigate the sociodemographic characteristics and experiences of patients who accessed their electronic health records (EHRs) through national patient portals in Norway, Sweden, Finland, and Estonia. METHODS: A cross-sectional web-based survey was distributed using the national online health portals. The target participants were patients who accessed the national patient portals at the start of 2022 and who were aged ≥15 years. The survey included a mixture of close-ended and free-text questions about participant sociodemographics, usability experience, experiences with health care and the EHR, reasons for reading health records online, experience with errors, omissions and offense, opinions about security and privacy, and the usefulness of portal functions. In this paper, we summarized the data on participant demographics, past experience with health care, and the patient portal through descriptive statistics. RESULTS: In total, 29,334 users completed the survey, of which 9503 (32.40%) were from Norway, 13,008 (44.35%) from Sweden, 4713 (16.07%) from Finland, and 2104 (7.17%) from Estonia. National samples were comparable according to reported gender, with about two-thirds identifying as women (19,904/29,302, 67.93%). Age distributions were similar across the countries, but Finland had older users while Estonia had younger users. The highest attained education and presence of health care education varied among the national samples. In all 4 countries, patients most commonly rated their health as "fair" (11,279/29,302, 38.48%). In Estonia, participants were more often inclined to rate their health positively, whereas Norway and Sweden had the highest proportion of negative health ratings. Across the whole sample, most patients received some care in the last 2 years (25,318/29,254, 86.55%). Mental health care was more common (6214/29,254, 21.24%) than oncological care (3664/29,254, 12.52%). Overall, most patients had accessed their health record "2 to 9 times" (11,546/29,306, 39.4%), with the most frequent users residing in Sweden, where about one-third of patients accessed it "more than 20 times" (4571/13,008, 35.14%). CONCLUSIONS: This is the first large-scale international survey to compare patient users' sociodemographics and experiences with accessing their EHRs. Although the countries are in close geographic proximity and demonstrate similar advancements in giving their residents online records access, patient users in this survey differed. We will continue to investigate patients' experiences and opinions about national patient-accessible EHRs through focused analyses of the national and combined data sets from the NORDeHEALTH 2022 Patient Survey.

Preserving Decentralized EHR-s Integrity.

A blockchain and decentralized storage technology-based architecture is proposed to support the integrity of Electronic Health Records (EHRs).

A country-wide teledermatoscopy service in Estonia shows results comparable to those in experimental settings in management plan development and diagnostic accuracy: A retrospective database study.

Background: Teledermatoscopy accuracy has been examined in experimental settings and is recommended for primary care despite lacking real-world implementation evidence. A teledermatoscopy service has been provided in Estonia since 2013, where lesions are evaluated based on the patient's or general practitioner's suggestion. Objective: The management plan and diagnostic accuracy of a real-world store-and-forward teledermatoscopy service for melanoma diagnosis were evaluated. Methods: A retrospective study analyzed 4748 cases from 3403 patients using the service between October 16, 2017 and August 30, 2019 by matching country-wide databases. Management plan accuracy was calculated as the percentage of melanoma found that was managed correctly. Diagnostic accuracy parameters were sensitivity, specificity, and positive and negative predictive values. Results: Management plan accuracy for melanoma detection was 95.5% (95% CI, 77.2-99.9). Diagnostic accuracy showed a sensitivity of 90.48% (95% CI, 69.62-98.83) and a specificity of 92.57% (95% CI, 91.79-93.31). Limitations: Matching the lesions was limited to SNOMED CT location standard precision. Diagnostic accuracy was calculated based on a combination of diagnosis and management plan data. Conclusion: Teledermatoscopy for detecting and managing melanoma in real-world clinical practice displays results comparable with those in experimental setting studies.

Digital Health Data Capture with a Controlled Natural Language.

Written text has been the preferred medium for storing health data ever since Hippocrates, and the medical narrative is what enables a humanized clinical relationship. Can't we admit natural language as a user-accepted technology that has stood against the test of time? We have previously presented a controlled natural language as a human-computer interface for semantic data capture already at the point of care. Our computable language was driven by a linguistic interpretation of the conceptual model of the Systematized Nomenclature of Medicine - Clinical Terms (SNOMED CT). This paper presents an extension that allows the capture of measurement results with numerical values and units. We discuss the relation our method can have with emerging clinical information modelling.

Migration from HL7 CDA to FHIR in Infectious Disease System of Estonia.

On the basis of the Estonian National Health System, we propose a universal and reusable methodology for migrating HL7 CDA documents or their components.

A systematic literature review of AI-based digital decision support systems for post-traumatic stress disorder.

Objective: Over the last decade, an increase in research on medical decision support systems has been observed. However, compared to other disciplines, decision support systems in mental health are still in the minority, especially for rare diseases like post-traumatic stress disorder (PTSD). We aim to provide a comprehensive analysis of state-of-the-art digital decision support systems (DDSSs) for PTSD. Methods: Based on our systematic literature review of DDSSs for PTSD, we created an analytical framework using thematic analysis for feature extraction and quantitative analysis for the literature. Based on this framework, we extracted information around the medical domain of DDSSs, the data used, the technology used for data collection, user interaction, decision-making, user groups, validation, decision type and maturity level. Extracting data for all of these framework dimensions ensures consistency in our analysis and gives a holistic overview of DDSSs. Results: Research on DDSSs for PTSD is rare and primarily deals with the algorithmic part of DDSSs (n = 17). Only one DDSS was found to be a usable product. From a data perspective, mostly checklists or questionnaires were used (n = 9). While the median sample size of 151 was rather low, the average accuracy was 82%. Validation, excluding algorithmic accuracy (like user acceptance), was mostly neglected, as was an analysis concerning possible user groups. Conclusion: Based on a systematic literature review, we developed a framework covering all parts (medical domain, data used, technology used for data collection, user interaction, decision-making, user groups, validation, decision type and maturity level) of DDSSs. Our framework was then used to analyze DDSSs for post-traumatic stress disorder. We found that DDSSs are not ready-to-use products but are mostly algorithms based on secondary datasets. This shows that there is still a gap between technical possibilities and real-world clinical work.

An evaluation of e-prescribing at a national level.

OBJECTIVE: The aim of the article is to describe the process of implementing the Estonian national second generation electronic prescription service (e-prescription) and determine if the objectives set by the Estonian government were fulfilled. MATERIALS AND METHODS: The study presents an analysis of both retrospective and current data in the evaluation phase of a design research project. Sources include, among others, publicly available documents and previous evaluation studies. RESULTS: According to all of the major stakeholders, the Estonian e-prescription service has very high usability and user satisfaction scores have been high. There is only little empirical evidence available to confirm if the benefits aimed for in the creation of the service were achieved. From a public administration viewpoint, the implementation of e-prescription has led to potential efficiency gains. CONCLUSION: The Estonian second-generation e-prescription system is widely used by citizens, healthcare providers and administrators alike. However, there are gaps in measuring the impact of the service, especially with respect to time savings and enhanced healthcare quality. Additional functionalities will be key drivers in creating benefits for all stakeholders. Future nationwide e-health services should have a more rigorous evaluation process carried out during the design and implementation stages.

Economically affordable anatomical kidney phantom with calyxes for puncture and drainage training in interventional urology and radiology.

BACKGROUND: Trends in interventional radiology and urology training are orientated towards reducing costs and increasing efficiency. In order to comply with the trends, we propose training on inexpensive patient-specific kidney phantoms. PURPOSE: To develop a new kidney phantom for puncture and drainage training in interventional urology and radiology, and to evaluate their anatomical correctness and suitability for training compared to the traditional way of training on home-made phantoms. MATERIAL AND METHODS: A CASE STUDY FOR VALIDATION OF KIDNEY PHANTOMS WAS CONDUCTED WITH NINE RADIOLOGY STUDENTS DIVIDED INTO TWO GROUPS: one trained on standard home-made training phantom (n = 4) and the other on our kidney phantoms (n = 5). Another test phantom was used to evaluate the effectiveness of the training of the two groups. The tests were video recorded and analyzed. Duration of the procedure was used as the primary indicator of procedure's quality. Comparison tests were also conducted with professional radiologists. Anatomical correctness of the kidney phantom was evaluated by comparing the post mortem kidney scans with reconstructed models from CT scans. Subjective feedback was also collected from the participants. Wider use of kidney phantoms was analyzed. RESULTS: The average volumetric difference between post mortem kidney scans and reconstructed CT kidney models was 4.70 ± 3.25%. All five students practicing on the kidney phantom improved their performance and the results were almost equal to the results of the professional radiologist while in the other group two students out of four trained on standard home-made training phantoms failed to improve their performance. However, the small number of test subjects prevents us from drawing general conclusions about the efficiency of the new practice. The kidney phantoms were found usable also for nephrostomy catheter placement training under fluoroscopy. CONCLUSION: The feedback from radiologists showed that the anatomically correct features of the phantom is an added value for the training and has a potential to increase the quality of minimally invasive procedures in general.

A comparison of approaches to providing patients access to summary care records across old and new europe: an exploration of facilitators and barriers to implementation.

Providing online access to their medical records should empower patients. National health services in Estonia, France, and England introduced systems to provide online access to summary health data. The Estonian patient access called "Patient Portal" shares similarities with the French system "Dossier Médical Personnel" (DMP). Both are patient-controlled records. The English system "Summary Care Record" (SCR) provided access to patients through "HealthSpace," though has now been replaced by encouraging access to GP records. Denmark and Ireland also provided access rights to patients so they could view their records. Romania, Croatia, and Greece do not have national systems. The aim of this study is to compare adoption and uptake of patient access to summary data. The Estonian record was used by 3.6% of the population, the French one by 0.5% and the English system is due for closure with only 0.01% signing up for the most comprehensive access. Few countries across Europe have adopted patient access to summary health data at a national level, and where introduced, medical records have been accessed by less than 5% of the population.

Low cost anatomically realistic renal biopsy phantoms for interventional radiology trainees.

This paper describes manufacturing of economically affordable renal biopsy phantoms for radiology residents and practicing radiologists. We reconstructed a realistic 3-dimensional patient-specific kidney model from CT data, manufactured an organ mould and casted the kidney phantoms. Using gelatin gel materials with calibrated parameters allowed making phantoms with realistic mechanical, ultrasound and CT properties including various pathologies. The organ phantoms with cysts included were further casted into gelatin gel medium. They were validated by radiology residents in biopsy training and compared against self-made phantoms traditionally used in the curriculum of interventional radiology. The realism, durability, price and suitability for training were evaluated. The results showed that our phantoms are more realistic and easier to use than the traditional ones. Our proposed technology allows creating a low-cost (50$/kg) alternative to the pricy commercial training phantoms available today.

Understanding drug uptake and binding within targeted disease micro-environments in patients: a new tool for translational medicine.

BACKGROUND: For many common global diseases, such as cancer, diabetes, neurodegenerative and cardiovascular diseases there is an unmet need for diagnosing early indications of disease that could enable medical intervention and early treatment. The treatment of these diseases will require detailed knowledge of targeted pathways involved in disease pathogenesis but also the mode of drug actions at the biological location on these targets. Translational medicine is a new area of research where expert from different disciplines involved in basic science and clinical disciplines meet and join forces. Mode-of-drug-action mechanisms elucidation is key in the characterization of drugs that can relate to both efficacy and safety. METHODS: Matrix assisted laser desorption/ionization mass spectrometry imaging (MALDI-MSI) was used providing evidence into the fate (destinations and distributions) of administered drugs within tumor regions of lung compartments. RESULTS: We hereby present a pulmonary study in which we have isolated lung tissue after inhaled drug administration and then localized the drug within airway wall compartments. The histology also provides evidence of drug binding to smooth muscle cell microenvironments. We also identified lung tissue regions with tumor cell invasion in these COPD patients. CONCLUSIONS: The ultimate goal is to identify bridging comprehension that forms a knowledge base that can be used by society to develop a better treatment and medicine for patients. Our results demonstrated that robust imaging data could be generated confirming drug localization in pulmonary regions of COPD patients with tumor pathology. TRIAL REGISTRATION: Tallinn Medical Research Ethical Committee decision #1724, 18.06.2009.

Direct demonstration of tissue uptake of an inhaled drug: proof-of-principle study using matrix-assisted laser desorption ionization mass spectrometry imaging.

Drug therapy is often directed to specific organ and tissue compartments where the mode of action of the compound affects specifically targeted biological processes. However, the direct measurement of drug uptake in terms of a time kinetic and concentrations attained at the local sites has not been readily available as a clinical index for most drugs. A proof-of-principle study was conducted to test the utility of applying matrix-assisted laser desorption ionization-mass spectrometry imaging (MALDI-MSI) to demonstrate the qualitative distribution pattern of a locally administered drug within tissue sites of targeted action. Here we have measured the occurrence of an inhaled bronchodilator, the muscarinic receptor antagonist ipratropium, within human bronchial biopsies obtained by fiber optic bronchoscopy shortly after dosing exposure. Cryo-preserved biopsy samples from five subjects being evaluated for airway obstruction or potential tumor development were prepared as thin frozen sections. Samples coated with a MALDI matrix were analyzed by a MALDI LTQ Orbitrap XL mass spectrometer at large (100 µm) and small (30 µm) raster sizes. Our results demonstrate that ipratropium is rapidly absorbed into the airway wall. Ipratropium parent ion (m/z 332.332) and daughter ions (m/z 166.2 and 290.2) were coincidently partitioned within submucosal spaces containing targeted airway smooth muscle in four out of five subjects. The signal intensity of ipratropium fragment ions provided estimates that local drug concentrations between 3 and 80 nM were achieved within the airway wall. To our knowledge, this is the first reported study in applying MALDI-MSI to demonstrate the localization of a drug administered at therapeutic levels. The study highlights the potential benefit of MALDI-MSI to provide important measurements of drug efficacy in clinical settings.

Nationwide Health Data Management System: a novel approach for integrating biomarker measurements with comprehensive health records in large populations studies.

The nation-wide electronic health record database acts as an interoperable repository of health data obtained throughout citizen contacts with health care providers. This system improves accessibility for citizens and researchers to health data with the ability to assign context to disease development. In that system, individual patients who are members of the large population-based health database can be assessed as individuals or as a population in prospective studies of prospective diseases.

Images crossing borders: image and workflow sharing on multiple levels.

Digitalisation of medical data makes it possible to share images and workflows between related parties. In addition to linear data flow where healthcare professionals or patients are the information carriers, a new type of matrix of many-to-many connections is emerging. Implementation of shared workflow brings challenges of interoperability and legal clarity. Sharing images or workflows can be implemented on different levels with different challenges: inside the organisation, between organisations, across country borders, or between healthcare institutions and citizens. Interoperability issues vary according to the level of sharing and are either technical or semantic, including language. Legal uncertainty increases when crossing national borders. Teleradiology is regulated by multiple European Union (EU) directives and legal documents, which makes interpretation of the legal system complex. To achieve wider use of eHealth and teleradiology several strategic documents were published recently by the EU. Despite EU activities, responsibility for organising, providing and funding healthcare systems remains with the Member States. Therefore, the implementation of new solutions requires strong co-operation between radiologists, societies of radiology, healthcare administrators, politicians and relevant EU authorities. The aim of this article is to describe different dimensions of image and workflow sharing and to analyse legal acts concerning teleradiology in the EU.

Patient opportunities in the Estonian Electronic Health Record System.

Estonia is the first country which has implemented a nationwide electronic health record system and gives full access to its citizens. This provides new opportunities to citizens, healthcare providers and e-health developers. Combining health data, citizens' self added health and welfare data, decision support and possibilities of service oriented architecture of the Estonian Electronic Health Record System - a new era of e-health services can begin.

Cross-border teleradiology-experience from two international teleradiology projects.

Teleradiology aims to even radiologists' workload, ensure on-call services, reduce waiting lists, consult other specialists and cut costs. Cross-border teleradiology widens this scope beyond the country borders. However, the new service should not reduce the quality of radiology. Quality and trust are key factors in establishment of teleradiology. Additionally there are organizational, technical, legal, security and linguistic issues influencing the service. Herein, we have used experiences from two partially European Union funded telemedicine projects to evaluate factors affecting cross-border teleradiology. Clinical partners from Czech Republic, Denmark, Estonia, Finland, Lithuania and the Netherlands went through 649 radiology test cases in two different teleradiology projects to build trust and agree about the report structure. Technical set-up was established using secure Internet data transfer, streaming technology, integration of workflows and creating structured reporting tool to overcome language barriers. The biggest barrier to overcome in cross-border teleradiology was the language issue. Establishment of the service was technically and semantically successful but limited to knee and hip X-ray examinations only because the structured reporting tool did not cover any other anatomical regions yet. Special attention has to be paid to clinical quality and trust between partners in cross-border teleradiology. Our experience shows that it is achievable. Legal, security and financial aspects are not covered in this paper because today they differ country by country. There is however an European Union level harmonization process started to enable cross-border eHealth in general.

Pervasive access to images and data--the use of computing grids and mobile/wireless devices across healthcare enterprises.

Emerging technologies are transforming the workflows in healthcare enterprises. Computing grids and handheld mobile/wireless devices are providing clinicians with enterprise-wide access to all patient data and analysis tools on a pervasive basis. In this paper, emerging technologies are presented that provide computing grids and streaming-based access to image and data management functions, and system architectures that enable pervasive computing on a cost-effective basis. Finally, the implications of such technologies are investigated regarding the positive impacts on clinical workflows.