Predicting substance use disorder using long-term attention deficit hyperactivity disorder medication records in Truven

Machine learning in attention-deficit/hyperactivity disorder: new approaches toward understanding the neural mechanisms

Attention-deficit/hyperactivity disorder (ADHD) is a highly prevalent and heterogeneous neurodevelopmental disorder in children and has a high chance of persisting in adulthood. The development of individualized, efficient, and reliable treatment strategies is limited by the lack of understanding of the underlying neural mechanisms. Diverging and inconsistent findings from existing studies suggest that ADHD may be simultaneously associated with multivariate factors across cognitive, genetic, and biological domains. Machine learning algorithms are more capable of detecting complex interactions between multiple variables than conventional statistical methods. Here we present a narrative review of the existing machine learning studies that have contributed to understanding mechanisms underlying ADHD with a focus on behavioral and neurocognitive problems, neurobiological measures including genetic data, structural magnetic resonance imaging (MRI), task-based and resting-state functional MRI (fMRI), electroencephalogram, and functional near-infrared spectroscopy, and prevention and treatment strategies. Implications of machine learning models in ADHD research are discussed. Although increasing evidence suggests that machine learning has potential in studying ADHD, extra precautions are still required when designing machine learning strategies considering the limitations of interpretability and generalization.

Identifying Opioid Use Disorder from Longitudinal Healthcare Data using a Multi-stream Transformer

Opioid Use Disorder (OUD) is a public health crisis costing the US billions of dollars annually in healthcare, lost workplace productivity, and crime. Analyzing longitudinal healthcare data is critical in addressing many real-world problems in healthcare. Leveraging the real-world longitudinal healthcare data, we propose a novel multi-stream transformer model called MUPOD for OUD identification. MUPOD is designed to simultaneously analyze multiple types of healthcare data streams, such as medications and diagnoses, by attending to segments within and across these data streams. Our model tested on the data from 392,492 patients with long-term back pain problems showed significantly better performance than the traditional models and recently developed deep learning models.

Integrating data science into the translational science research spectrum: A substance use disorder case study

The availability of large healthcare datasets offers the opportunity for researchers to navigate the traditional clinical and translational science research stages in a nonlinear manner. In particular, data scientists can harness the power of large healthcare datasets to bridge from preclinical discoveries (T0) directly to assessing population-level health impact (T4). A successful bridge from T0 to T4 does not bypass the other stages entirely; rather, effective team science makes a direct progression from T0 to T4 impactful by incorporating the perspectives of researchers from every stage of the clinical and translational science research spectrum. In this exemplar, we demonstrate how effective team science overcame challenges and, ultimately, ensured success when a diverse team of researchers worked together, using healthcare big data to test population-level substance use disorder (SUD) hypotheses generated from preclinical rodent studies. This project, called Advancing Substance use disorder Knowledge using Big Data (ASK Big Data), highlights the critical roles that data science expertise and effective team science play in quickly translating preclinical research into public health impact.

Deep Learning Prediction of Mild Cognitive Impairment using Electronic Health Records

About 44.4 million people have been diagnosed with dementia worldwide, and it is estimated that this number will be almost tripled by 2050. Predicting mild cognitive impairment (MCI), an intermediate state between normal cognition and dementia and an important risk factor for the development of dementia is crucial in aging populations. MCI is formally determined by health professionals through a comprehensive cognitive evaluation, together with a clinical examination, medical history and often the input of an informant (an individual that know the patient very well). However, this is not routinely performed in primary care visits, and could result in a significant delay in diagnosis. In this study, we used deep learning and machine learning techniques to predict the progression from cognitively unimpaired to MCI and also to analyze the potential for patient clustering using routinely-collected electronic health records (EHRs). Our analysis of EHRs indicates that temporal characteristics of patient data incorporated in a deep learning model provides increased power in predicting MCI.