Model and Strategy for Predicting and Discovering Drug-Drug Interactions.

Taking several medications at the same time is an increasingly common phenomenon in our society. The combination of drugs is certainly not without risk of potentially dangerous interactions. Taking into account all possible interactions is a very complex task as it is not yet known what all possible interactions between drugs and their types are. Machine learning based models have been developed to help with this task. However, the output of these models is not structured enough to be integrated in a clinical reasoning process on interactions. In this work, we propose a clinically relevant and technically feasible model and strategy for drug interactions.

Translating the Observational Medical Outcomes Partnership - Common Data Model (OMOP-CDM) Electronic Health Records to an OWL Ontology.

The heterogeneity of electronic health records model is a major problem: it is necessary to gather data from various models for clinical research, but also for clinical decision support. The Observational Medical Outcomes Partnership - Common Data Model (OMOP-CDM) has emerged as a standard model for structuring health records populated from various other sources. This model is proposed as a relational database schema. However, in the field of decision support, formal ontologies are commonly used. In this paper, we propose a translation of OMOP-CDM into an ontology, and we explore the utility of the semantic web for structuring EHR in a clinical decision support perspective, and the use of the SPARQL language for querying health records. The resulting ontology is available online.

A Qualitative Method for Learning Medical Expert Reasoning.

The aim of this paper is to propose a qualitative method for learning a model that represents the closest possible experts reasoning and strategies to provide recommendations of antibiotics. The learned model contains an integrity constraint and a preference formula. The former indicates the features that an antibiotic should have to be recommended. The later indicates the rank of recommendation of an antibiotic.

Speak-PIM, Towards a Framework for the Automatic Detection of Potentially Inappropriate Prescriptions.

Potentially inappropriate medications (PIMs) have adverse health consequences, particularly in elderly patients. Various explicit criteria have been developed to detect PIMs. However, it is difficult to apply these criteria without the help of an electronic decision support tool. Programming these tools can be very complex. Indeed, for computer scientists it is difficult to understand medical issues and for clinicians it is difficult to program in a computer programming language. In this work we present Speak-PIM, a framework for formalizing the PIM's rules. Speak-PIM is based on a very simple semantics which is suitable for the declaration of PIMs without embarking on all the complexity of description logic or computer languages. It aims to offer an efficient collaboration between the computer scientists and clinicians.

Decision-support systems for managing polypharmacy in the elderly: A scoping review.

Polypharmacy, the consuming of more than five drugs, is a public health problem. It can lead to many interactions and adverse drug reactions and is very expensive. Therapeutic guidelines for managing polypharmacy in the elderly have been issued, but are highly complex, limiting their use. Decision-support systems have therefore been developed to automate the execution of these guidelines, or to provide information about drugs adapted to the context of polypharmacy. These systems differ widely in terms of their technical design, knowledge sources and evaluation methods. We present here a scoping review of electronic systems for supporting the management, by healthcare providers, of polypharmacy in elderly patients. Most existing reviews have focused mainly on evaluation results, whereas the present review also describes the technical design of these systems and the methodologies for developing and evaluating them. A systematic bibliographic search identified 19 systems differing considerably in terms of their technical design (rule-based systems, documentary approach, mixed); outputs (textual report, alerts and/or visual approaches); and evaluations (impact on clinical practices, impact on patient outcomes, efficiency and/or user satisfaction). The evaluations performed are minimal (among all the systems identified, only one system has been evaluated according to all the criteria mentioned above) and no machine learning systems and/or conflict management systems were retrieved. This review highlights the need to develop new methodologies, combining various approaches for decision support system in polypharmacy.

ABiMed: Towards an Innovative Clinical Decision Support System for Medication Reviews and Polypharmacy Management.

Polypharmacy in elderly is a public health problem with both clinical (increase of adverse drug events) and economic issues. One solution is medication review, a structured assessment of patients' drug orders by the pharmacist for optimizing the therapy. However, this task is tedious, cognitively complex and error-prone, and only a few clinical decision support systems have been proposed for supporting it. Existing systems are either rule-based systems implementing guidelines, or documentary systems presenting drug knowledge. In this paper, we present the ABiMed research project, and, through literature reviews and brainstorming, we identified five candidate innovations for a decision support system for medication review: patient data transfer from GP to pharmacists, use of semantic technologies, association of rule-based and documentary approaches, use of machine learning, and a two-way discussion between pharmacist and GP after the medication review.

Encouraging Behavior Changes and Preventing Cardiovascular Diseases Using the Prevent Connect Mobile Health App: Conception and Evaluation of App Quality.

BACKGROUND: Cardiovascular diseases are a major cause of death worldwide. Mobile health apps could help in preventing cardiovascular diseases by improving modifiable risk factors such as eating habits, physical activity levels, and alcohol or tobacco consumption. OBJECTIVE: The aim of this study was to design a mobile health app, Prevent Connect, and to assess its quality for (1) assessing patient behavior for 4 cardiovascular risk factors (unhealthy eating, sedentary lifestyle, alcohol, and tobacco consumption) and (2) suggesting personalized recommendations and mobile health interventions for risky behaviors. METHODS: The knowledge base of the app is based on French national recommendations for healthy eating, physical activity, and limiting alcohol and tobacco consumption. It contains a list of patient behaviors and related personalized recommendations and digital health interventions. The interface was designed according to usability principles. Its quality was assessed by a panel of 52 users in a 5-step process: completion of the demographic form, visualization of a short presentation of the app, testing of the app, completion of the user version of the Mobile App Rating Scale (uMARS), and an open group discussion. RESULTS: This app assesses patient behaviors through specific questionnaires about 4 risk factors (unhealthy eating, sedentary lifestyle, alcohol, and tobacco consumption) and suggests personalized recommendations and digital health interventions for improving behavior. The app was deemed to be of good quality, with a mean uMARS quality score of 4 on a 5-point Likert scale. The functionality and information content of the app were particularly appreciated, with a mean uMARS score above 4. Almost all the study participants appreciated the navigation system and found the app easy to use. More than three-quarters of the study participants found the app content relevant, concise, and comprehensive. The aesthetics and the engagement of the app were also appreciated (uMARS score, 3.7). Overall, 80% (42/52) of the study participants declared that the app helped them to become aware of the importance of addressing health behavior, and 65% (34/52) said that the app helped motivate them to change lifestyle habits. CONCLUSIONS: The app assessed the risky behaviors of the patients and delivered personalized recommendations and digital health interventions for multiple risk factors. The quality of the app was considered to be good, but the impact of the app on behavior changes is yet to be demonstrated and will be assessed in further studies.

Visual Comparison of Guidelines: Method and Application to Potentially Inappropriate Medication Lists.

Therapeutic guidelines developed by experts are essential tools for improving therapy and drug prescription. Several guidelines often exist that target the same patient, from different organizations and countries. The case of lists for the detection of potentially inappropriate medications (PIMs) is an example which illustrates how these guidelines can be varied and multiple. In order to have an overview to the divergences and similarities between different lists of PIMs, we propose a visual method to compare PIMs lists, based on set visualization, and we apply it to 5 guidelines.

Influence of Connected Health Interventions for Adherence to Cardiovascular Disease Prevention: A Scoping Review.

BACKGROUND: Recent health care developments include connected health interventions to improve chronic disease management and/or promote actions reducing aggravating risk factors for conditions such as cardiovascular diseases. Adherence is one of the main challenges for ensuring the correct use of connected health interventions over time. OBJECTIVE: This scoping review deals with the connected health interventions used in interventional studies, describing the ways in which these interventions and their functions effectively help patients to deal with cardiovascular risk factors over time, in their own environments. The objective is to acquire knowledge and highlight current trends in this field, which is currently both productive and immature. METHODS: A structured literature review was constructed from Medline-indexed journals in PubMed. We established inclusion criteria relating to three dimensions (cardiovascular risk factors, connected health interventions, and level of adherence). Our initial search yielded 98 articles; 78 were retained after screening on the basis of title and abstract, 49 articles underwent full-text screening, and 24 were finally retained for the analysis, according to preestablished inclusion criteria. We excluded studies of invasive interventions and studies not dealing with digital health. We extracted a description of the connected health interventions from data for the population or end users. RESULTS: We performed a synthetic analysis of outcomes, based on the distribution of bibliometrics, and identified several connected health interventions and main characteristics affecting adherence. Our analysis focused on three types of user action: to read, to do, and to connect. Finally, we extracted current trends in characteristics: connect, adherence, and influence. CONCLUSION: Connected health interventions for prevention are unlikely to affect outcomes significantly unless other characteristics and user preferences are considered. Future studies should aim to determine which connected health design combinations are the most effective for supporting long-term changes in behavior and for preventing cardiovascular disease risks.

Visualization of Drug Interactions for Supporting Medication Review.

The number of elderly patients with multimorbidities has considerably increased since recent years. These patients are often polymedicated and at higher risk of safety incidents due to polypharmacy. To reduce polypharmacy, one solution is Medication review (MR). MR aimed at optimizing drug treatments, is unfortunately not very frequent in practice. Indeed, consulting the properties of 5-20 drugs in parallel is a cognitively complex task. It is therefore necessary to develop software for supporting MR. The existing tools only list alerts concerning drugs and their interactions. The objective of our work is to facilitate the pharmacist's access to the medical knowledge necessary for drug interactions. Using visual analytics, we propose an interactive tool that synthesizes information on drug interactions. It shows an overview of drug treatment and make it visually accessible by the pharmacist to facilitate MR.

An Approach Based on Preference Learning for Identifying Experts Reasoning in Antibiotic Treatment.

Clinical Practice Guidelines (CPGs) aim at improving the quality of health care by providing standardized best practices for diagnosis and treatment. However, physicians have difficulties to understand the (often implicit) rationale underlying expert recommendations. The aim of this paper is to propose an approach based on preference learning for building a model that is closest to the reasoning of experts to provide recommendations. We apply this method to antibiotherapy in primary care. The preference model was learned from the recommendations and from a database describing the domain.

Decision Support System for Selection of e-Health Interventions.

The main goal of this work was to design a decision support system for effective personalized cardiovascular risk prevention: i) to identify behavioral groups associated with clinical risk factors, ii) to provide recommendations associated with the objective to be achieved and iii) to determine the decision-making rules assigning each group to the type of mobile health intervention conveying the most appropriate prevention messages, to help patients to achieve attainable goals. The system is based on an existing data prediction model taking into account specific risky behaviors, clinical risk factors and social status, and it is embedded in a new e-health application. The system is operational. The next step will be the design of a large study to assess improvements in patient adherence to prevention messages through e-health interventions selected by the application.

Bank of Digital Filters for Identification of Combined Drug Products in a French Public Database.

Drugs information systems, prescription support softwares, and drug decision support systems need to reason on drug properties. Combined pharmaceutical products need to be considered specifically because they may require a specific processing. Hence, they also need to be identified to automate the population of databases with up-to-date property values. We defined a set of digital filters designed for the identification of antibiotics in a public database. Four different filters are proposed, to be combined to extract the relevant information. Evaluation was conducted to combine filters and retrieve information about rand combined antibiotics with success. However, information provided in the structured files of the French drug database is limited; information provided in the HTML files suffers from a lack of quality. Hence, reuse of this data and this information should be performed very cautiously.

Identifying Experts Reasoning in Antibiotic Treatment with Preference Learning.

The aim of this paper is to propose an approach based on preference learning for building a model that represents the closest possible experts reasoning to provide recommendations. We apply this method to antibiotherapy in primary care. The preference model was learned from the recommendations and from a database describing the domain.

Explainable decision support through the learning and visualization of preferences from a formal ontology of antibiotic treatments.

The aim of eXplainable Artificial Intelligence (XAI) is to design intelligent systems that can explain their predictions or recommendations to humans. Such systems are particularly desirable for therapeutic decision support, because physicians need to understand rcommendations to have confidence in their application and to adapt them if required, e.g. in case of patient contraindication. We propose here an explainable and visual approach for decision support in antibiotic treatment, based on an ontology. There were three steps to our method. We first generated a tabular dataset from the ontology, containing features defined on various domains and n-ary features. A preference model was then learned from patient profiles, antibiotic features and expert recommendations found in clinical practice guidelines. This model made the implicit rationale of the expert explicit, including the way in which missing data was treated. We then visualized the preference model and its application to all antibiotics available on the market for a given clinical situation, using rainbow boxes, a recently developed technique for set visualization. The resulting preference model had an error rate of 3.5% on the learning data, and 5.2% on test data (10-fold validation). These findings suggest that our system can help physicians to prescribe antibiotics correctly, even for clinical situations not present in the guidelines (e.g. due to allergies or contraindications for the recommended treatment).

AntibioGame®: A serious game for teaching medical students about antibiotic use.

INTRODUCTION: Measures for controlling antimicrobial resistance are urgently required. We describe here AntibioGame®, a serious game for improving the training of medical students in antibiotic use in primary care. OBJECTIVE: We aimed to design a serious game for antibiotics teaching and to evaluate its usability and playability by medical students. METHODS: We used various gamification techniques (e.g. use of mascots, avatars, rewards, leader board) and cartoon graphics in the design of AntibioGame®. This game implements clinical case templates built from a list of learning goals defined by a medical team through an analysis of clinical practice guidelines. The game was evaluated by asking medical students to rate their satisfaction and the usability and playability of the game on an electronic form and through group discussions. The electronic form was derived from the MEEGA + scale, a five-point Likert scale including 32 items for assessing both usability and playability. RESULTS: AntibioGame® is a case-based game in which students play the role of a doctor meeting patients in consultation and helping other health professionals to solve their problems, as in real life. The scenarios are realistic and cover situations frequently encountered in primary care. The 57 medical students enrolled found the game attractive, usable, fun, and appropriate for learning. Game quality was considered "good" (score = 60 on the MEEGA + scale). All the students said they would recommend the game, 96 % liked it and 81 % would use it for revision. CONCLUSION: AntibioGame® is a promising tool for improving knowledge in antibiotic prescription that could easily be included in multifaceted programs for training medical students.

Helping GPs to extrapolate guideline recommendations to patients for whom there are no explicit recommendations, through the visualization of drug properties. The example of AntibioHelp® in bacterial diseases.

INTRODUCTION: Clinical decision support systems (CDSS) implementing clinical practice guidelines (CPGs) have 2 main limitations: they target only patients for whom CPGs provide explicit recommendations, and their rationale may be difficult to understand. These 2 limitations result in poor CDSS adoption. We designed AntibioHelp® as a CDSS for antibiotic treatment. It displays the recommended and nonrecommended antibiotics, together with their properties, weighted by degree of importance as outlined in the CPGs. The aim of this study was to determine whether AntibioHelp® could increase the confidence of general practitioners (GPs) in CPG recommendations and help them to extrapolate guidelines to patients for whom CPGs provide no explicit recommendations. MATERIALS AND METHODS: We carried out a 2-stage crossover study in which GPs responded to clinical cases using CPG recommendations either alone or with explanations displayed through AntibioHelp®. We compared error rates, confidence levels, and response times. RESULTS: We included 64 GPs. When no explicit recommendation existed for a particular situation, AntibioHelp® significantly decreased the error rate (-41%, P value = 6x10-13), and significantly increased GP confidence (+8%, P value = .02). This CDSS was considered to be usable by GPs (SUS score = 64), despite a longer interaction time (+9-22 seconds). By contrast, AntibioHelp® had no significant effect if there was an explicit recommendation. DISCUSSION/CONCLUSION: The visualization of weighted antibiotic properties helps GPs to extrapolate recommendations to patients for whom CPGs provide no explicit recommendations. It also increases GP confidence in their prescriptions for these patients. Further evaluations are required to determine the impact of AntibioHelp® on antibiotic prescriptions in real clinical practice.

Using preference learning for detecting inconsistencies in clinical practice guidelines: Methods and application to antibiotherapy.

Clinical practice guidelines provide evidence-based recommendations. However, many problems are reported, such as contradictions and inconsistencies. For example, guidelines recommend sulfamethoxazole/trimethoprim in child sinusitis, but they also state that there is a high bacteria resistance in this context. In this paper, we propose a method for the semi-automatic detection of inconsistencies in guidelines using preference learning, and we apply this method to antibiotherapy in primary care. The preference model was learned from the recommendations and from a knowledge base describing the domain. We successfully built a generic model suitable for all infectious diseases and patient profiles. This model includes both preferences and necessary features. It allowed the detection of 106 candidate inconsistencies which were analyzed by a medical expert. 55 inconsistencies were validated. We showed that therapeutic strategies of guidelines in antibiotherapy can be formalized by a preference model. In conclusion, we proposed an original approach, based on preferences, for modeling clinical guidelines. This model could be used in future clinical decision support systems for helping physicians to prescribe antibiotics.

Building Bayesian Networks from Causal Rules.

Bayesian Networks (BNs) are often used for designing diagnosis decision support systems. They are a well-established method for reasoning under uncertainty and making inferences. But, eliciting the probabilities can be tedious and time-consuming especially in medical domain where variables are often related by qualitative terms rather than probabilities. The goal of this paper is to propose a method for eliciting the probabilities required in BNs by using and transforming causal rules which are often used in medicine. The method consists in first constructing the structure of BNs by reporting medical expert's knowledge in the form of causal rules, and then constructing the parameters of the BNs by transforming the terms used for qualified causal rules into probabilities. Example is given in obesity domain. Further works are needed to reinforce our method like the consideration of circular causal rules.

Decision System Integrating Preferences to Support Sleep Staging.

Scoring sleep stages can be considered as a classification problem. Once the whole recording segmented into 30-seconds epochs, features, extracted from raw signals, are typically injected into machine learning algorithms in order to build a model able to assign a sleep stage, trying to mimic what experts have done on the training set. Such approaches ignore the advances in sleep medicine, in which guidelines have been published by the AASM, providing definitions and rules that should be followed to score sleep stages. In addition, these approaches are not able to solve conflict situations, in which criteria of different sleep stages are met. This work proposes a novel approach based on AASM guidelines. Rules are formalized integrating, for some of them, preferences allowing to support decision in conflict situations. Applied to a doubtful epoch, our approach has taken the appropriate decision.