Development of a machine learning-based predictive model for prediction of success or failure of medical management for benign prostatic hyperplasia.

OBJECTIVE: To develop a novel predictive model for identifying patients who will and will not respond to the medical management of benign prostatic hyperplasia (BPH). METHODS: Using data from the Medical Therapy of Prostatic Symptoms (MTOPS) study, several models were constructed using an initial data set of 2172 patients with BPH who were treated with doxazosin (Group 1), finasteride (Group 2), and combination therapy (Group 3). K-fold stratified cross-validation was performed on each group, Within each group, feature selection and dimensionality reduction using nonnegative matrix factorization (NMF) were performed based on the training data, before several machine learning algorithms were tested; the most accurate models, boosted support vector machines (SVMs), being selected for further refinement. The area under the receiver operating curve (AUC) was calculated and used to determine the optimal operating points. Patients were classified as treatment failures or responders, based on whether they fell below or above the AUC threshold for each group and for the whole data set. RESULTS: For the entire cohort, the AUC for the boosted SVM model was 0.698. For patients in Group 1, the AUC was 0.729, for Group 2, the AUC was 0.719, and for Group 3, the AUC was 0.698. CONCLUSION: Using MTOPS data, we were able to develop a prediction model with an acceptable rate of discrimination of medical management success for BPH.

α-Satellite: An AI-Driven System and Benchmark Datasets for Dynamic COVID-19 Risk Assessment in the United States.

The fast evolving and deadly outbreak of coronavirus disease (COVID-19) has posed grand challenges to human society. To slow the spread of virus infections and better respond for community mitigation, by advancing capabilities of artificial intelligence (AI) and leveraging the large-scale and up-to-date data generated from heterogeneous sources (e.g., disease related data, demographic, mobility and social media data), in this work, we propose and develop an AI-driven system (named α-Satellite), as an initial offering, to provide dynamic COVID-19 risk assessment in the United States. More specifically, given a point of interest (POI), the system will automatically provide risk indices associated with it in a hierarchical manner (e.g., state, county, POI) to enable people to select appropriate actions for protection while minimizing disruptions to daily life. To comprehensively evaluate our system for dynamic COVID-19 risk assessment, we first conduct a set of empirical studies; and then we validate it based on a real-world dataset consisting of 5,060 annotated POIs, which achieves the area of under curve (AUC) of 0.9202. As of June 18, 2020, α-Satellite has had 56,980 users. Based on the feedback from its large-scale users, we perform further analysis and have three key findings: i) people from more severe regions (i.e., with larger numbers of COVID-19 cases) have stronger interests using our system to assist with actionable information; ii) users are more concerned about their nearby areas in terms of COVID-19 risks; iii) the user feedback about their perceptions towards COVID-19 risks of their query POIs indicate the challenge of public concerns about the safety versus its negative effects on society and the economy. Our system and generated datasets have been made publicly accessible via our website.

Development and Validation of a Machine Learning Algorithm for Predicting Response to Anticholinergic Medications for Overactive Bladder Syndrome.

OBJECTIVE: To develop and externally validate a prediction model for anticholinergic response in patients with overactive bladder (OAB). METHODS: A machine learning model to predict the likelihood of anticholinergic treatment failure was constructed using a retrospective data set (n=559) of female patients with OAB who were treated with anticholinergic medications between January 2010 and December 2017. Treatment failure was defined as less than 50% improvement in frequency, urgency, incontinence episodes, and nocturia, and the patient's subjective impression of symptomatic relief. Patients were stratified by age (younger than 40 years, 40-60 years, and older than 60 years), and number of previously failed medications. K-fold stratified cross-validation was performed on each stratum using machine learning algorithms. Of these, the random forest model was the most accurate. This model was refined using internal cross validation within each stratum. The area under the curve (AUC) was calculated for each stratum and used to identify the optimal operating points for prediction of treatment failure. The random forest model was then externally validated using a prospectively collected data set (n=82) of women treated with anticholinergic medications at a different clinical site between January 2018 and December 2018. RESULTS: The global accuracy of the final model was 80.3% (95% CI 79.1-81.3), and the AUC was 0.77 (95% CI 0.74-0.79). Using the external validation data set, the model's sensitivity and specificity was 80.4% (95% CI 66.5-89.7%) and 77.4% (95% CI 58.6-89.7%), respectively. The model performed best in women aged younger than 40 years (AUC 0.84, 95% CI 0.81-0.84) and worst in women aged older than 60 years who had previously failed medication (AUC 0.71, 95% CI 0.67-0.75). CONCLUSION: Our externally validated machine learning prediction model can predict anticholinergic treatment failure during the standard 3-month treatment trial period with greater than 80% accuracy. The model can be accessed at https://oabweb.herokuapp.com/app/pre/.