Learning to Classify Medical Discharge Summaries According to ICD-9.

CONTEXT: We present a post-hoc approach to improve the recall of ICD classification. METHOD: The proposed method can use any classifier as a backbone and aims to calibrate the number of codes returned per document. We test our approach on a new stratified split of the MIMIC-III dataset. RESULTS: When returning 18 codes on average per document we obtain a recall that is 20% better than a classic classification approach.

Patient Electronic Health Record as Temporal Graphs for Health Monitoring.

Machine learning methods are becoming increasingly popular to anticipate critical risks in patients under surveillance reducing the burden on caregivers. In this paper, we propose an original modeling that benefits of recent developments in Graph Convolutional Networks: a patient's journey is seen as a graph, where each node is an event and temporal proximities are represented by weighted directed edges. We evaluated this model to predict death at 24 hours on a real dataset and successfully compared our results with the state of the art.

Discovering NDM-1 inhibitors using molecular substructure embeddings representations.

NDM-1 (New-Delhi-Metallo-ß-lactamase-1) is an enzyme developed by bacteria that is implicated in bacteria resistance to almost all known antibiotics. In this study, we deliver a new, curated NDM-1 bioactivities database, along with a set of unifying rules for managing different activity properties and inconsistencies. We define the activity classification problem in terms of Multiple Instance Learning, employing embeddings corresponding to molecular substructures and present an ensemble ranking and classification framework, relaying on a k-fold Cross Validation method employing a per fold hyper-parameter optimization procedure, showing promising generalization ability. The MIL paradigm displayed an improvement up to 45.7 %, in terms of Balanced Accuracy, in comparison to the classical Machine Learning paradigm. Moreover, we investigate different compact molecular representations, based on atomic or bi-atomic substructures. Finally, we scanned the Drugbank for strongly active compounds and we present the top-15 ranked compounds.

VERTIGo: A Visual Platform for Querying and Exploring Large Multilayer Networks.

Many real world data can be modeled by a graph with a set of nodes interconnected to each other by multiple relationships. Such a rich graph is called multilayer graph or network. Providing useful visualization tools to support the query process for such graphs is challenging. Although many approaches have addressed the visual query construction, few efforts have been done to provide a contextualized exploration of query results and suggestion strategies to refine the original query. This is due to several issues such as i) the size of the graphs ii) the large number of retrieved results and iii) the way they can be organized to facilitate their exploration. In this article, we present VERTIGo, a novel visual platform to query, explore and support the analysis of large multilayer graphs. VERTIGo provides coordinated views to navigate and explore the large set of retrieved results at different granularity levels. In addition, the proposed system supports the refinement of the query by visual suggestions to guide the user through the exploration process. Two examples and a user study demonstrate how VERTIGo can be used to perform visual analysis (query, exploration, and suggestion) on real world multilayer networks.

Mining microarray data to predict the histological grade of a breast cancer.

BACKGROUND: The aim of this study was to develop an original method to extract sets of relevant molecular biomarkers (gene sequences) that can be used for class prediction and can be included as prognostic and predictive tools. MATERIALS AND METHODS: The method is based on sequential patterns used as features for class prediction. We applied it to classify breast cancer tumors according to their histological grade. RESULTS: We obtained very good recall and precision for grades 1 and 3 tumors, but, like other authors, our results were less satisfactory for grade 2 tumors. CONCLUSIONS: We demonstrated the interest of sequential patterns for class prediction of microarrays and we now have the material to use them for prognostic and predictive applications.

Prediction of In-Hospital Mortality from Administrative Data: A Sequential Pattern Mining Approach.

Study of trajectory of care is attractive for predicting medical outcome. Models based on machine learning (ML) techniques have proven their efficiency for sequence prediction modeling compared to other models. Introducing pattern mining techniques contributed to reduce model complexity. In this respect, we explored methods for medical events' prediction based on the extraction of sets of relevant event sequences of a national hospital discharge database. It is illustrated to predict the risk of in-hospital mortality in acute coronary syndrome (ACS). We mined sequential patterns from the French Hospital Discharge Database. We compared several predictive models using a text string distance to measure the similarity between patients' patterns of care. We computed combinations of similarity measurements and ML models commonly used. A Support Vector Machine model coupled with edit-based distance appeared as the most effective model. Indeed discrimination ranged from 0.71 to 0.99, together with a good overall accuracy. Thus, sequential patterns mining appear motivating for event prediction in medical settings as described here for ACS.

Hospital healthcare flows: A longitudinal clustering approach of acute coronary syndrome in women over 45 years.

Acute coronary syndrome (ACS) in women is a growing public health issue and a death leading cause. We explored whether the hospital healthcare trajectory was characterizable using a longitudinal clustering approach in women with ACS. From the 2009-2014 French nationwide hospital database, we extracted spatio-temporal patterns in ACS patient trajectories, by replacing the spatiality by their hospitalization cause. We used these patterns to characterize hospital healthcare flows in a visualization tool. We clustered these trajectories with kmlShape to identify time gap and tariff profiles. ACS hospital healthcare flows have three key categories: Angina pectoris, Myocardial Infarction or Ischemia. Elderly flows were more complex. Time gap profiles showed that readmissions were closer together as time goes by. Tariff profiles were different according to age and initial event. Our approach might be applied to monitoring other chronic diseases. Further work is needed to integrate these results into a medical decision-making tool.

Discovering novelty in sequential patterns: application for analysis of microarray data on Alzheimer disease.

UNLABELLED: Analyzing microarrays data is still a great challenge since existing methods produce huge amounts of useless results. We propose a new method called NoDisco for discovering novelties in gene sequences obtained by applying data-mining techniques to microarray data. METHOD: We identify popular genes, which are often cited in the literature, and innovative genes, which are linked to the popular genes in the sequences but are not mentioned in the literature. We also identify popular and innovative sequences containing these genes. Biologists can thus select interesting sequences from the two sets and obtain the k-best documents. RESULTS: We show the efficiency of this method by applying it on real data used to decipher the mechanisms underlying Alzheimer disease. CONCLUSION: The first selection of sequences based on popularity and innovation help experts focus on relevant sequences while the top-k documents help them understand the sequences.

Expertise in French health forums.

More and more health websites hire medical experts (physicians, medical students, experienced volunteers, etc.) and indicate explicitly their medical role in order to notify that they provide high-quality answers. However, medical experts may participate in forum discussions even when their role is not officially indicated. Detecting posts written by medical experts facilitates the quick access to posts that have more chances of being correct and informative. The main objective of this work is to learn classification models that can be used to detect posts written by medical experts in any health forum discussions. Two French health forums have been used to discover the best features and methods for this text categorization task. The obtained results confirm that models learned on appropriate websites may be used efficiently on other websites (more than 98% of F1-measure has been obtained using a Random Forest classifier). A study of misclassified posts highlights the participation of medical experts in forum discussions even if their role is not explicitly indicated.

MultiStream: A Multiresolution Streamgraph Approach to Explore Hierarchical Time Series.

Multiple time series are a set of multiple quantitative variables occurring at the same interval. They are present in many domains such as medicine, finance, and manufacturing for analytical purposes. In recent years, streamgraph visualization (evolved from ThemeRiver) has been widely used for representing temporal evolution patterns in multiple time series. However, streamgraph as well as ThemeRiver suffer from scalability problems when dealing with several time series. To solve this problem, multiple time series can be organized into a hierarchical structure where individual time series are grouped hierarchically according to their proximity. In this paper, we present a new streamgraph-based approach to convey the hierarchical structure of multiple time series to facilitate the exploration and comparisons of temporal evolution. Based on a focus+context technique, our method allows time series exploration at different granularities (e.g., from overview to details). To illustrate our approach, two usage examples are presented.

PaFloChar: An Innovating Approach to Characterise Patient Flows in Myocardial Infarction.

A better knowledge of patient flows would improve decision making in health planning. In this article, we propose a method to characterise patients flows and also to highlight profiles of care pathways considering times and costs. From medico-administrative data, we extracted spatio-temporal patterns. Then, we clustered time between hospitalisations and cost trajectories in order to identify profiles of change over time. This approach may support renewed management strategies.

E-Patient Reputation in Health Forums.

Online health forums are increasingly used by patients to get information and help related to their health. However, information reliability in these forums is unfortunately not always guaranteed. Obviously, consequences of self-diagnosis may be severe on the patient's health if measures are taken without consulting a doctor. Many works on trust issues related to social media have been proposed, but most of them mainly focus only on the structure part of the social network (number of posts, number of likes, etc.). In the case of online health forums, a lot of trust and distrust is expressed inside the posted messages and cannot be inferred by only considering the structure. In this study, we rather suggest inferring the user's trustworthiness from the replies he receives in the forum. The proposed method is divided into three main steps: First, the recipient(s) of each post must be identified. Next, the trust or distrust expressed in these posts is evaluated. Finally, the user's reputation is computed by aggregating all the posts he received. Conducted experiments using a manually annotated corpus are encouraging.

Assisting e-patients in an Ask the Doctor Service.

Ask the doctor services are personalized forums allowing patients to ask questions directly to doctors. Usually, patients must choose the most appropriate category for their question among lots of categories to be redirected to the most relevant physician. However, manual selection is tedious and error prone activity. In this work we propose to assist the patients in this task by recommending a short list of most appropriate categories.

Gender-related differences in the management of hypertension by cardiologists: the PARITE study.

BACKGROUND: Several studies have shown gender differences in the management of cardiovascular risk factors and diseases. Whether the management of hypertension by cardiologists in France differs according to patient gender has not been fully investigated. AIMS: The main objective of this cross-sectional, multicentre study was to examine the management according to gender of hypertensive patients by office-based cardiologists in France. METHODS: Cardiologists were asked to include consecutively two men and two women attending a routine consultation for essential hypertension. Therapeutic management was evaluated by comparing cardiovascular investigations in the preceding 6 months and hypertension control according to gender and the patients' global cardiovascular risk. RESULTS: Overall, data from 3440 adult patients (53% men) referred to 654 cardiologists were analysed. Hypertension was uncontrolled in 76% of both men and women and 69% were at high global cardiovascular risk (75% of men, 62% of women; P<0.001). Significantly fewer cardiovascular investigations had been performed in the preceding 6 months in women (22.6% vs 44.2% in men; P<0.001). The treatment regimen was changed by the cardiologist in approximately 50% of patients regardless of gender or global cardiovascular risk. CONCLUSIONS: The PARITE study shows that in French office-based cardiology practice, the antihypertensive regimen is adjusted as often in female as in male patients. However, the results suggest that there is room for improvement in the investigation of cardiovascular disease in women. Healthcare providers could be encouraged to implement established guidelines on the prevention of cardiovascular disease in women.