Reasoning with large language models for medical question answering.

OBJECTIVES: To investigate approaches of reasoning with large language models (LLMs) and to propose a new prompting approach, ensemble reasoning, to improve medical question answering performance with refined reasoning and reduced inconsistency. MATERIALS AND METHODS: We used multiple choice questions from the USMLE Sample Exam question files on 2 closed-source commercial and 1 open-source clinical LLM to evaluate our proposed approach ensemble reasoning. RESULTS: On GPT-3.5 turbo and Med42-70B, our proposed ensemble reasoning approach outperformed zero-shot chain-of-thought with self-consistency on Steps 1, 2, and 3 questions (+3.44%, +4.00%, and +2.54%) and (2.3%, 5.00%, and 4.15%), respectively. With GPT-4 turbo, there were mixed results with ensemble reasoning again outperforming zero-shot chain-of-thought with self-consistency on Step 1 questions (+1.15%). In all cases, the results demonstrated improved consistency of responses with our approach. A qualitative analysis of the reasoning from the model demonstrated that the ensemble reasoning approach produces correct and helpful reasoning. CONCLUSION: The proposed iterative ensemble reasoning has the potential to improve the performance of LLMs in medical question answering tasks, particularly with the less powerful LLMs like GPT-3.5 turbo and Med42-70B, which may suggest that this is a promising approach for LLMs with lower capabilities. Additionally, the findings show that our approach helps to refine the reasoning generated by the LLM and thereby improve consistency even with the more powerful GPT-4 turbo. We also identify the potential and need for human-artificial intelligence teaming to improve the reasoning beyond the limits of the model.

Disparities in Retinal Vein Occlusion Presentation and Initiation of Anti-VEGF Therapy: An Academy IRIS® Registry Analysis.

OBJECTIVE: Investigate disparities in retinal vein occlusion (RVO) presentation and initiation of anti-VEGF treatment. DESIGN: Retrospective cohort study. SUBJECTS: Patients in the American Academy of Ophthalmology IRIS® (Intelligent Research in Sight) Registry database (2015-2021) with branch or central RVO and macular edema (ME). METHODS: The association of demographic characteristics and presenting visual acuity (VA) with anti-VEGF treatment initiation were quantified using multivariable logistic regression. MAIN OUTCOME MEASURES: Treatment with ≥ 1 anti-VEGF injection within 12 months after RVO diagnosis. RESULTS: A total of 304 558 eligible patients with RVO and ME were identified. Age at presentation varied by race, ethnicity, sex, and RVO type (all P values < 0.001). Within the first year after RVO presentation, 192 602 (63.2%) patients received ≥ 1 anti-VEGF injection. In a multivariable regression model adjusting for relevant covariates, female (vs. male) patients had lower odds of receiving injections (odds ratio [OR], 0.95; 95% confidence interval [CI], 0.93-0.96; P < 0.0001) as did Black/African American (vs. White) patients (OR, 0.90; 95% CI, 0.88-0.92; P < 0.0001) and Asian (vs. White) patients (OR, 0.95; 95% CI, 0.91-0.99; P = 0.02), whereas older patients (vs. patients aged < 51 years) had higher odds (61-70 years: OR, 1.20; 95% CI, 1.16-1.24; 71-80 years: OR, 1.20; 95% CI, 1.16-1.24; > 80 years: OR, 1.15; 95% CI, 1.11-1.18; all P values < 0.0001). Hispanic (vs. non-Hispanic) patients had a small increased odds of treatment initiation (OR, 1.08; 95% CI, 1.04-1.11; P < 0.0001). Results were similar in the subset of 226 143 patients with VA data. In this subset, patients with presenting VA < 20/40 to 20/200 were most frequently treated in the first year after diagnosis (∼ 70%) and patients with light perception/no light perception (LP-NLP) vision or VA of 20/20 or better were treated least frequently (36.9% and 41.9%, respectively). CONCLUSIONS: In this large national clinical registry, 37% of RVO patients with ME had no anti-VEGF treatment documented in the first year after diagnosis. Black/African American, Asian, and female patients and patients with VA of LP-NLP were least likely to receive treatment. Awareness of this undertreatment and these disparities highlight the need for initiatives to ensure all RVO patients receive timely anti-VEGF injections for optimized visual outcomes. FINANCIAL DISCLOSURES: Proprietary or commercial disclosure may be found in the Footnotes and Disclosures at the end of this article.

Frequent temporal patterns of physiological and biological biomarkers and their evolution in sepsis.

Sepsis is one of the most challenging health conditions worldwide, with relatively high incidence and mortality rates. It is shown that preventing sepsis is the key to avoid potentially irreversible organ dysfunction. However, data-driven early identification of sepsis is challenging as sepsis shares signs and symptoms with other health conditions. This paper adopts a temporal pattern mining approach to identify frequent temporal and evolving patterns of physiological and biological biomarkers in sepsis patients. We show that using these frequent patterns as features for classifying sepsis and non-sepsis patients can improve the prediction accuracy and performance up to 7%. Most of the temporal modeling approaches adopted in the sepsis literature are based on deep learning methods. Although these approaches produce high accuracy, they generally have limited model explainability and interpretability. Using the adopted methods in this study, we could identify the most important features contributing to the patients' sepsis incidence, such as fluctuations in platelet, lactate, and creatinine, or evolution of patterns including renal and metabolic organ systems, and consequently, enhance the findings' clinical interpretability.

Examining Public Awareness of Ageist Terms on Twitter: Content Analysis.

Background: The World Health Organization, the Centers for Disease Control and Prevention, and the Gerontological Society of America have made efforts to raise awareness on ageist language and propose appropriate terms to denote the older adult population. The COVID-19 pandemic and older adults' vulnerability to the disease have perpetuated hostile ageist discourse on social media. This is an opportune time to understand the prevalence and use of ageist language and discuss the ways forward. Objective: This study aimed to understand the prevalence and situated use of ageist terms on Twitter. Methods: We collected 60.32 million tweets between March and July 2020 containing terms related to COVID-19. We then conducted a mixed methods study comprising a content analysis and a descriptive quantitative analysis. Results: A total of 58,930 tweets contained the ageist terms "old people" or "elderly." The more appropriate term "older adult" was found in 11,328 tweets. Twitter users used ageist terms (eg, "old people" and "elderly") to criticize ageist messages (17/60, 28%), showing a lack of understanding of appropriate terms to describe older adults. Highly hostile ageist content against older adults came from tweets that contained the derogatory terms "old people" (22/30, 73%) or "elderly" (13/30, 43%). Conclusions: The public discourse observed on Twitter shows a continued lack of understanding of appropriate terms to use when referring to older adults. Effort is needed to eliminate the perpetuation of ageist messages that challenge healthy aging. Our study highlights the need to inform the public about appropriate language use and ageism.

The IHI Rochester Report 2022 on Healthcare Informatics Research: Resuming After the CoViD-19.

In 2020, the CoViD-19 pandemic spread worldwide in an unexpected way and suddenly modified many life issues, including social habits, social relationships, teaching modalities, and more. Such changes were also observable in many different healthcare and medical contexts. Moreover, the CoViD-19 pandemic acted as a stress test for many research endeavors, and revealed some limitations, especially in contexts where research results had an immediate impact on the social and healthcare habits of millions of people. As a result, the research community is called to perform a deep analysis of the steps already taken, and to re-think steps for the near and far future to capitalize on the lessons learned due to the pandemic. In this direction, on June 09th-11th, 2022, a group of twelve healthcare informatics researchers met in Rochester, MN, USA. This meeting was initiated by the Institute for Healthcare Informatics-IHI, and hosted by the Mayo Clinic. The goal of the meeting was to discuss and propose a research agenda for biomedical and health informatics for the next decade, in light of the changes and the lessons learned from the CoViD-19 pandemic. This article reports the main topics discussed and the conclusions reached. The intended readers of this paper, besides the biomedical and health informatics research community, are all those stakeholders in academia, industry, and government, who could benefit from the new research findings in biomedical and health informatics research. Indeed, research directions and social and policy implications are the main focus of the research agenda we propose, according to three levels: the care of individuals, the healthcare system view, and the population view.

Mental health conditions and unsafe driving behaviors: A naturalistic driving study on ADHD and depression.

INTRODUCTION: Road injuries remain a persistent public health concern across the world. The task of driving is complicated by mental health conditions, which may affect drivers' executive functioning and cognitive resource allocation. This study examines whether attention-deficit/hyperactivity disorder (ADHD) and depression are associated with unsafe driving behaviors. METHOD: Generalized linear mixed models were employed to estimate the association of self-reported ADHD and depression with 18 unsafe driving behavior types found prior to at-fault crashes and near-crashes using a large-scale naturalistic driving dataset. Driver demographics, cognitive traits, environmental factors, and driver random effects were included to reduce confounding and biases. RESULTS: Controlling for other covariates, people with self-reported ADHD were more likely to have performed improper braking or stopping (OR = 4.89, 95% CI 1.82-13.17) prior to an at-fault crash or near-crash, while those with self-reported depression did not have a significant association with any unsafe driving behavior. CONCLUSIONS: After accounting for demographic, cognitive, and environmental covariates, individuals with ADHD and depression were not prone to purposefully aggressive or reckless driving. Instead, drivers with self-reported ADHD may unintentionally execute unsafe driving behaviors in particular driving scenarios that require a high level of cognitive judgment. PRACTICAL APPLICATIONS: These findings can inform the curriculum design of driver's education programs that help learners with mental health conditions gain practice in certain road scenarios, for example, more practice on preemptively reducing speed instead of making sudden brakes and smooth turning on curved roads for students with ADHD. Furthermore, specific advanced driver assistance systems may prove particularly helpful for drivers with ADHD, such as detection of leading objects and curve speed warning.

Explainable Artificial Intelligence for Predictive Modeling in Healthcare.

The principle behind artificial intelligence is mimicking human intelligence in the way that it can perform tasks, recognize patterns, or predict outcomes through learning from the acquired data of various sources. Artificial intelligence and machine learning algorithms have been widely used in autonomous driving, recommender systems in electronic commerce and social media, fintech, natural language understanding, and question answering systems. Artificial intelligence is also gradually changing the landscape of healthcare research (Yu et al. in Biomed Eng 2:719-731, 25). The rule-based approach that relied on the curation of medical knowledge and the construction of robust decision rules had drawn significant attention in diagnosing diseases and clinical decision support since half a century ago. In recent years, machine learning algorithms such as deep learning that can account for complex interactions between features is shown to be promising in predictive modeling in healthcare (Deo in Circulation 132:1920-1930, 26). Although many of these artificial intelligence and machine learning algorithms can achieve remarkably high performance, it is often difficult to be completely adopted in practical clinical environments due to the lack of explainability in some of these algorithms. Explainable artificial intelligence (XAI) is emerging to assist in the communication of internal decisions, behavior, and actions to health care professionals. Through explaining the prediction outcomes, XAI gains the trust of the clinicians as they may learn how to apply the predictive modeling in practical situations instead of blindly following the predictions. There are still many scenarios to explore how to make XAI effective in clinical settings due to the complexity of medical knowledge.

How are different sources of distraction associated with at-fault crashes among drivers of different age gender groups?

INTRODUCTION: Distracted driving has been well researched, however the comparison between different age-gender groups on the impact of distracted driving has not been explored. Most crash analysis research does not distinguish driver responsibility, so the role that distractions has in at-fault crashes is unknown. Without distinguishing at-fault crashes from all-cause crashes, distracted driving's detrimental effects could be underestimated. OBJECTIVE: This study aims to systematically assess the risk of at-fault crashes associated with different sources of distraction among six groups by driver age (Teens 16-19, Adults 20-64, Seniors 65+) and gender. METHODS: Crashes where a study participant was deemed at fault were identified using human expert annotated variables from the Strategic Highway Research Program 2 (SHRP2) Naturalistic Driving Study dataset. Generalized linear mixed models were performed to assess the adjusted odds ratios of 10 distraction types associated with the at-fault crashes while controlling for environmental factors. RESULTS: The main findings are (1) The highest contributing distraction types in at-fault crashes were In-Cabin Objects, Mobile Device, External Scenes, and In-Vehicle Information Systems (IVIS) as indicated by their influence on multiple age-gender groups and the magnitude of odds ratios; (2) Teens and adults were more distraction-prone than seniors, although seniors had the greatest at-fault crash risks associated with In-Cabin Objects, Mobile Device, and IVIS; (3) Distractions impacted females and males similarly; (4) At-fault crashes were more likely to have the significant distraction types present than all-cause crashes. CONCLUSION: This study adds to the limited literature on at-fault crashes particularly as it explores the role of driver demographics and distracted driving. Analyzing the risks of distracted driving by age-gender group shows that specific activities can be riskier for a certain population. The effects of distractions may be overlooked without fault determination. Distractions by external scenes, in-vehicle technologies, and in-cabin objects should not be overlooked, in addition to mobile device use.

Determining the risk of driver-at-fault events associated with common distraction types using naturalistic driving data.

INTRODUCTION: Studies thus far have focused on automobile accidents that involve driver distraction. However, it is hard to discern whether distraction played a role if fault designation is missing because an accident could be caused by an unexpected external event over which the driver has no control. This study seeks to determine the effect of distraction in driver-at-fault events. METHOD: Two generalized linear mixed models, one with at-fault safety critical events (SCE) and the other with all-cause SCEs as the outcomes, were developed to compare the odds associated with common distraction types using data from the SHRP2 naturalistic driving study. RESULTS: Adjusting for environment and driver variation, 6 of 10 common distraction types significantly increased the risk of at-fault SCEs by 20-1330%. The three most hazardous sources of distraction were handling in-cabin objects (OR = 14.3), mobile device use (OR = 2.4), and external distraction (OR = 1.8). Mobile device use and external distraction were also among the most commonly occurring distraction types (10.1% and 11.0%, respectively). CONCLUSIONS: Focusing on at-fault events improves our understanding of the role of distraction in potentially avoidable automobile accidents. The in-cabin distraction that requires eye-hand coordination presents the most danger to drivers' ability in maintaining fault-free, safe driving. Practical Applications: The high risk of at-fault SCEs associated with in-cabin distraction should motivate the smart design of the interior and in-vehicle information system that requires less visual attention and manual effort.

Topics and Sentiments of Public Concerns Regarding COVID-19 Vaccines: Social Media Trend Analysis.

BACKGROUND: As a number of vaccines for COVID-19 are given emergency use authorization by local health agencies and are being administered in multiple countries, it is crucial to gain public trust in these vaccines to ensure herd immunity through vaccination. One way to gauge public sentiment regarding vaccines for the goal of increasing vaccination rates is by analyzing social media such as Twitter. OBJECTIVE: The goal of this research was to understand public sentiment toward COVID-19 vaccines by analyzing discussions about the vaccines on social media for a period of 60 days when the vaccines were started in the United States. Using the combination of topic detection and sentiment analysis, we identified different types of concerns regarding vaccines that were expressed by different groups of the public on social media. METHODS: To better understand public sentiment, we collected tweets for exactly 60 days starting from December 16, 2020 that contained hashtags or keywords related to COVID-19 vaccines. We detected and analyzed different topics of discussion of these tweets as well as their emotional content. Vaccine topics were identified by nonnegative matrix factorization, and emotional content was identified using the Valence Aware Dictionary and sEntiment Reasoner sentiment analysis library as well as by using sentence bidirectional encoder representations from transformer embeddings and comparing the embedding to different emotions using cosine similarity. RESULTS: After removing all duplicates and retweets, 7,948,886 tweets were collected during the 60-day time period. Topic modeling resulted in 50 topics; of those, we selected 12 topics with the highest volume of tweets for analysis. Administration and access to vaccines were some of the major concerns of the public. Additionally, we classified the tweets in each topic into 1 of the 5 emotions and found fear to be the leading emotion in the tweets, followed by joy. CONCLUSIONS: This research focused not only on negative emotions that may have led to vaccine hesitancy but also on positive emotions toward the vaccine. By identifying both positive and negative emotions, we were able to identify the public's response to the vaccines overall and to news events related to the vaccines. These results are useful for developing plans for disseminating authoritative health information and for better communication to build understanding and trust.

The Exchange of Informational Support in Online Health Communities at the Onset of the COVID-19 Pandemic: Content Analysis.

BACKGROUND: Online health communities (OHCs) provide social support for ongoing health-related problems. COVID-19, the disease caused by SARS-CoV-2, has been an acute and substantial stressor worldwide. The disease and its impact, especially in the beginning phases, left many people with questions about the nature, treatment, and prevention of COVID-19. Unlike typical chronic ailments discussed on OHCs, which are more established, COVID-19, at least at the onset of the pandemic, is distinct in that it lacks a consensus of clinical diagnosis and an existing community foundation. OBJECTIVE: The study aims to investigate a newly formed OHC for COVID-19 to determine the topics and types of information exchange as well as the sources of information this community referenced during the early phases of the COVID-19 pandemic in the United States. METHODS: A total of 357 posts from a COVID-19 OHC on the MedHelp platform were annotated according to an open-coding process. Participants' engagement patterns, topics of posts, and sources of information were quantified. RESULTS: Participants who offered informational support had a significantly higher percentage of responding more than once than those seeking information (P<.001). Among the information-seeking topics, symptoms and public health practice and psychological impacts were the most frequently discussed, with 26% (17/65) and 15% (10/65) of posts, respectively. Most informational support was expressed through feedback/opinion (181/220, 82.3%). Additionally, the most frequently referenced source of information was news outlets/websites, at 55% (11/20). Governmental websites were referenced less frequently. CONCLUSIONS: The trends of this community could be useful in prioritizing public health responses to address the most common questions asked by the public during crisis communication and in identifying which venue of communication is most effective in reaching a public audience during such times.

Proximity of Cellular and Physiological Response Failures in Sepsis.

Sepsis is a devastating multi-stage health condition with a high mortality rate. Its complexity, prevalence, and dependency of its outcomes on early detection have attracted substantial attention from data science and machine learning communities. Previous studies rely on individual cellular and physiological responses representing organ system failures to predict health outcomes or the onset of different sepsis stages. However, it is known that organ systems' failures and dynamics are not independent events. In this study, we identify the dependency patterns of significant proximate sepsis-related failures of cellular and physiological responses using data from 12,223 adult patients hospitalized between July 2013 and December 2015. The results show that proximate failures of cellular and physiological responses create better feature sets for outcome prediction than individual responses. Our findings reveal the few significant proximate failures that play the major roles in predicting patients' outcomes. This study's results can be simply translated into clinical practices and inform the prediction and improvement of patients' conditions and outcomes.

Identifying Self-Management Support Needs for Pregnant Women With Opioid Misuse in Online Health Communities: Mixed Methods Analysis of Web Posts.

BACKGROUND: The current opioid crisis in the United States impacts broad population groups, including pregnant women. Opioid use during pregnancy can affect the health and wellness of both mothers and their infants. Understanding women's efforts to self-manage opioid use or misuse in pregnancy is needed to identify intervention points for improving maternal outcomes. OBJECTIVE: This study aims to identify the characteristics of women in an online health community (OHC) with opioid use or misuse during pregnancy and the self-management support needs of these mothers. METHODS: A total of 200 web posts by pregnant women with opioid use participating in an OHC were double coded. Concepts and their thematic connections were identified through an inductive process until theoretical saturation was reached. Statistical tests were performed to identify patterns. RESULTS: The majority of pregnant women (150/200, 75.0%) in the OHC exhibited signs of misuse, and 62.5% (125/200) of the participants were either contemplating or pursuing dosage reduction. Self-managed withdrawal was more common (P<.001) than professional treatment among the population. A total of 5 themes of self-management support needs were identified as women sought information about the potential adverse effects of gestational opioid use, protocols for self-managed withdrawal, pain management safety during pregnancy, hospital policies and legal procedures related to child protection, and strategies for navigating offline support systems. In addition, 58.5% (117/200) of the pregnant women expressed negative emotions, of whom only 10.2% (12/117) sought to address their emotional needs with the help of the OHC. CONCLUSIONS: OHCs provide vital self-management support for pregnant women with opioid use or misuse. Women pursuing self-managed dosage reduction are prone to misinformation and repeated relapses, which can result in extreme measures to avoid testing positive for drug use at labor. The study findings provide evidence for public policy considerations, including universal screening of substance use for pregnant women, emphasis on treatment rather than legal punishment, and further expansion of the Drug Addiction Treatment Act waiver training program. The improvement of web-based platforms that can organize geo-relevant information, dispense clinically validated withdrawal schedules, and offer structured peer support is envisioned for harm reduction among pregnant women who opt for self-management of opioid misuse.

The Impact of driver distraction and secondary tasks with and without other co-occurring driving behaviors on the level of road traffic crashes.

Driving safety is typically affected by concurrent non-driving tasks. These activities might negatively impact the trips' outcome and cause near-crash or crash incidents and accidents. The crashes impose a tremendous social and economic cost to society and might affect the involving individuals' quality of life. As it stands, road injuries are ranked among top-ten leading causes of death by the World Health Organization. Distracted driving is defined as an attention diversion of the driver toward a competing activity. It was shown in numerous studies that distracted driving increase the probability of near-crash or crash events. By leveraging the statistical power of the large SHRP2 naturalistic data, we are able to quantify the preponderance of specific distractions during daily trips and confirm the causality factor of an ubiquitous non-driving task in the crash event. We show that, except for phone usage which happens more frequently in near-crash and crash categories than in baseline trips, both distracted driving and secondary tasks occur almost uniformly in different types of trips. In this study, we investigate the impact of the co-occurrence of distracted driving with other driving behaviors and secondary tasks. It is found that the co-occurrence of distracted driving with other driving behaviors or secondary tasks increase the chance of near-crash and crash events. This study's findings can inform the design and development of more precise and reliable driving assistance and warning systems.

What Are People Concerned About During the Pandemic? Detecting Evolving Topics about COVID-19 from Twitter.

With the novel coronavirus (COVID-19) pandemic affecting the lives of the citizens of over 200 countries, there is a need for policy makers and clinicians to understand public sentiment and track the spread of the disease. One of the sources for gaining valuable insight into public sentiment is through social media. This study aims to extract this insight by producing a list of the most discussed topics regarding COVID-19 on Twitter every week and monitoring the evolution of topics from week to week. This research will propose two topic mining that can handle a large-scale dataset-rolling online non-negative matrix factorization (Rolling-ONMF) and sliding online non-negative matrix factorization (Sliding-ONMF)-and compare the insights produced by both techniques. Each algorithm produces 425 topics over the course of 17 weeks. However, topics that have not evolved from one week to the next beyond a certain evolution threshold are consolidated into a single topic. Since the topics produced by the Rolling-ONMF algorithm each week depend on the topics from the previous week, we find that the Sliding-ONMF algorithm produces more varied topics each week; however, the topics produced by the Rolling-ONMF algorithm contain keywords that appear more consistent with each other when reviewing the terms manually. We also observe that the Sliding-ONMF algorithm is able to capture events that have shorter time frames rather than ones that last throughout many months while the Rolling-ONMF algorithm detects more general themes due to a higher average evolution score which leads to more topic consolidation. We have also conducted a qualitative analysis and grouped the detected topics into themes. A number of important themes such as government policy, economic crisis, COVID-19-related updates, COVID-19-related events, prevention, vaccines and treatments, and COVID-19 testing are identified. These reflected the concerns related to the pandemic expressed in social media.

Connectedness Based on Shared Engagement Predicts Remote Biochemically Verified Quit Status Within Smoking Cessation Treatment Groups on Facebook.

INTRODUCTION: Engagement with online content and online social network integration are associated with smoking behavior change, but less is known about social dynamics of shared engagement between participants in group-based social media interventions. METHODS: Participants were 251 young adult smokers aged 18 to 25 assigned to one of 29 secret Facebook groups tailored to their readiness to quit smoking ("pre-contemplation," "contemplation," and "preparation"). Groups varied in size and were randomly assigned to receive monetary incentives for engagement. All groups received daily posts for 90 days and were assessed for remote biochemically verified smoking abstinence at the end of the intervention. Across 29 groups, we examined associations between group features (group size, incentive condition, readiness to quit) with how connected members were within the group based on shared engagement with the same content (measured by density). At the individual level, we examined associations between 7-day biochemically verified smoking abstinence and how connected an individual was within the group (measured by degree centrality). RESULTS: After adjusting for comment volume, being in a contemplation group (vs. pre-contemplation group) was associated with a decrease in comment-based density. Individual degree centrality was significantly associated with biochemically verified smoking abstinence for both comments and likes. CONCLUSIONS: Future group-based social media interventions for smoking cessation may want to focus on promoting connected engagement between participants, rather than simply quantity of engagement. IMPLICATIONS: Participants in a smoking cessation intervention delivered through Facebook groups were more likely to have biochemically verified smoking abstinence if they were more connected to the rest of the group via shared engagement. Promoting shared engagement between participants may be more likely to promote behavior change than volume of engagement alone.

Imputation of Missing Data in Electronic Health Records Based on Patients' Similarities.

Using electronic health records (EHR) as the source of data for mining and analysis of different health conditions has become an increasingly common approach. However, due to irregular observation times and other uncertainties inherent in medical settings, the EHR data sets suffer from a large number of missing values. Most of the traditional data mining and machine learning approaches are designed to operate on complete data. In this paper, we propose a novel imputation method for missing data to facilitate using these approaches for the analysis of EHR data. The imputation is based on a set of interpatient, multivariate similarities among patients. For a missing data point in a patient's lab results during his/her intensive care unit stay, the method ranks other patients based on their similarities with the ego patient in terms of lab values, then the missing value is estimated as a weighted average of the known values of the same laboratory test from other patients, considering their similarities as weights. A comparison of the estimated values by the proposed method with values estimated by several common and state-of-the-are methods, such as MICE and 3D-MICE, shows that the proposed method outperforms them and produces promising results.

Mining heterogeneous network for drug repositioning using phenotypic information extracted from social media and pharmaceutical databases.

Drug repositioning has drawn significant attention for drug development in pharmaceutical research and industry, because of its advantages in cost and time compared with the de novo drug development. The availability of biomedical databases and online health-related information, as well as the high-performance computing, empowers the development of computational drug repositioning methods. In this work, we developed a systematic approach that identifies repositioning drugs based on heterogeneous network mining using both pharmaceutical databases (PharmGKB and SIDER) and online health community (MedHelp). By utilizing adverse drug reactions (ADRs) as the intermediate, we constructed a heterogeneous health network containing drugs, diseases, and ADRs, and developed path-based heterogeneous network mining approaches for drug repositioning. Additionally, we investigated on how the data sources affect the performance on drug repositioning. Experiment results showed that combining both PharmKGB and MedHelp identified 479 repositioning drugs, which are more than the repositioning drugs discovered by other alternatives. In addition, 31% of the 479 of the discovered repositioning drugs were supported by evidence from PubMed.

Drug Repositioning to Accelerate Drug Development Using Social Media Data: Computational Study on Parkinson Disease.

BACKGROUND: Due to the high cost and low success rate in new drug development, systematic drug repositioning methods are exploited to find new indications for existing drugs. OBJECTIVE: We sought to propose a new computational drug repositioning method to identify repositioning drugs for Parkinson disease (PD). METHODS: We developed a novel heterogeneous network mining repositioning method that constructed a 3-layer network of disease, drug, and adverse drug reaction and involved user-generated data from online health communities to identify potential candidate drugs for PD. RESULTS: We identified 44 non-Parkinson drugs by using the proposed approach, with data collected from both pharmaceutical databases and online health communities. Based on the further literature analysis, we found literature evidence for 28 drugs. CONCLUSIONS: In summary, the proposed heterogeneous network mining repositioning approach is promising for identifying repositioning candidates for PD. It shows that adverse drug reactions are potential intermediaries to reveal relationships between disease and drug.