The specific aim of this study is to develop machine learning models as a clinical approach for personalized treatment of osteoporosis. The model performance on outcome prediction was compared between four machine learning algorithms. Retrospective, electronic clinical data for patients with suspected or confirmed osteoporosis treated at Wan Fang Hospital between 2011 to 2018 were used as inputs for building the following predictive machine learning models,i.e., artificial neural network (ANN), random forest (RF), support vector machine (SVM) and logistic regression (LR) models. The predicted outcome was defined as an increase/decrease in T-score after treatment. A genetic algorithm was employed to select relevant variables as input features for each model; the leave-one-out method was applied for model building and internal validation. The model with best performance was selected by a separate set of testing. Area under the receiver operating characteristic curve, accuracy, precision, sensitivity and F1 score were calculated to evaluate model performance. Main analysis for all the patients with subclinical or confirmed osteoporosis and subgroup analysis for the patients with confirmed osteoporosis (T score < -2.5) were carried out in this study. A genetic algorithm was employed to select 12 to 18 features from all 33 variables for the four models. No difference was found in accuracy (ANN, 71.7%; LR, 70.0%; RF, 75.0%; SVM, 66.7%), precision (ANN, 80.0%; LR, 59.3%; RF, 70.0%; SVM, 63.6%), and AUC (ANN, 0.709; LR, 0.731; RF, 0.719; SVM, 0.702) among the ANN, LR, RF and SVM models. Main analysis in performance revealed significant recall in the LR model, as compared to ANN and SVM model; while subgroup revealed significant recall in ANN model, compared to LR and SVM model. Machine learning-based models hold potential in forecasting the outcomes of treatment for osteoporosis via early initiation of first-line therapy for patients with subclinical disease; or a switch to second-line treatment for patients with a high risk of impending treatment failure. This convenient approach can assist clinicians in adjusting treatment tailored to individual patient for prevention of disease progression or ineffective therapy.

The integrase strand transfer inhibitor (INSTI)-containing regimens are currently considered as the first-line treatment of human immunodeficiency virus (HIV) infection. Although possessing a common mechanism of action to inhibit HIV integrase irreversibly to stop HIV replication cycle, the INSTIs, including raltegravir, elvitegravir, dolutegravir, and bictegravir, differ in pharmacokinetic characteristics. While raltegravir undergoes biotransformation by the UDP-glucuronosyltransferases (UGTs), elvitegravir is primarily metabolized by cytochrome P450 (CYP) 3A4 and co-formulated with cobicistat to increase its plasma exposure. The metabolism pathways of dolutegravir and bictegravir are similar, both including CYP3A and UGT1A1, and both agents are substrates to different drug transporters. Because of their differences in metabolism, INSTIs interact with other medications differently through CYP enzymes and transporters as inducers or inhibitors. These drug interactions may become an important consideration in the long-term clinical use because the life expectancy of people with HIV (PWH) approaches to that of the general population. Also, common geriatric challenges such as multimorbidity and polypharmacy have been increasingly recognized in PWH. This review provides a summary of pharmacokinetic interactions with INSTIs and future perspectives in implications of INSTI drug interactions.

Background: The Hospital Readmissions Reduction Program (HRRP) was introduced to reduce readmission rates among Medicare beneficiaries, however little is known about readmissions and costs for HRRP-targeted conditions in younger populations. The primary objective of this study was to examine readmission trends and costs for targeted conditions during policy implementation among younger and older adults in the U.S. Methods: We analyzed the Nationwide Readmission Database from January 2010 to September 2015 in younger (18–64 years) and older (≥65 years) patients with acute myocardial infarction (AMI), heart failure (HF), pneumonia, and acute exacerbations of chronic obstructive pulmonary disease (AECOPD). Pre- and post-HRRP periods were defined based on implementation of the policy for each condition. Readmission rates were evaluated using an interrupted time series with difference-in-difference analyses and hospital cost differences between early and late readmissions (≤30 vs. > 30 days) were evaluated using generalized linear models. Results: Overall, this study included 16,884,612 hospitalizations with 3,337,266 readmissions among all age groups and 5,977,177 hospitalizations with 1,104,940 readmissions in those aged 18–64 years. Readmission rates decreased in all conditions. In the HRRP announcement period, readmissions declined significantly for those aged 40–64 years for AMI (p < 0.0001) and HF (p = 0.003). Readmissions decreased significantly in the post-HRRP period for those aged 40–64 years at a slower rate for AMI (p = 0.003) and HF (p = 0.05). Readmission rates among younger patients (18–64 years) varied within all four targeted conditions in HRRP announcement and post-HRRP periods. Adjusted models showed a significantly higher readmission cost in those readmitted within 30 days among younger and older populations for AMI (p < 0.0001), HF (p < 0.0001), pneumonia (p < 0.0001), and AECOPD (p < 0.0001). Conclusion: Readmissions for targeted conditions decreased in the U.S. during the enactment of the HRRP policy and younger age groups (< 65 years) not targeted by the policy saw a mixed effect. Healthcare expenditures in younger and older populations were significantly higher for early readmissions with all targeted conditions. Further research is necessary evaluating total healthcare utilization including emergency department visits, observation units, and hospital readmissions in order to better understand the extent of the HRRP on U.S. healthcare.

Almost every facet of modern biomedical research involves artificial intelligence (AI). This ACCP commentary forecasts the role of AI in clinical pharmacy research and scholarship. The potential benefits/opportunities together with the limitations/challenges of AI are reviewed for stages of the scientific method including (1) developing the research question(s), study design, and execution; (2) data analysis; and (3) reporting and dissemination of clinical pharmacy research. Benefits and opportunities of AI in clinical pharmacy research include streamlining hypothesis generation and facilitating study design, overcoming limitations of traditional statistical analysis techniques, facilitating manuscript development and dissemination, and expediting peer review. Limitations and challenges of AI include the introduction of biases in subject recruitment; generation of false information, also known as “AI hallucinations”; concern of “black box” analyses that are difficult to validate; potential legal liabilities; lack of accountability; and the need for investigators to ensure the accuracy and integrity of AI‐generated content. In summary, rapid progress of AI capabilities has great potential to revolutionize and accelerate clinical pharmacy research and scholarship; however, it is also imperative to recognize and mitigate the challenges and limitations introduced by AI.

Objectives: To determine the impact of a pharmacist-led telephone outreach program among patients discharged from the emergency department (ED) to home. Study Design: We conducted a randomized controlled study from February to November 2019 at a tertiary care academic medical center. Methods: At ED discharge, participants were randomly assigned to usual care (controls) or usual care plus the pharmacist’s review (intervention group). Eligible individuals included those being discharged from the ED to home with 8 or more medications. A pharmacist telephoned patients in the intervention group within 48 to 96 hours after ED discharge. The medications in the patient’s record from the ED were compared with what the patient was taking at home. Discrepancies were communicated to the primary provider via fax or telephone. The primary outcome was overall health care utilization including unplanned hospital readmissions or ED visits within 30 days of discharge. The effect of the intervention on the number of acute events was analyzed using a Poisson regression model adjusting for relevant baseline characteristics. Results: Of 90 eligible participants, 45 patients each were in the intervention and control groups. A total of 26 patients (58%) in the intervention group were reached, and 56 interventions were provided by the pharmacists. There was no significant difference between groups for overall health care utilization (adjusted risk ratio [aRR], 1.01; 95% CI, 0.50-2.06; ,P, = .96), hospitalizations (aRR, 0.20; 95% CI, 0.02-2.18; ,P, = .19), and ED visits (aRR, 1.24; 95% CI, 0.56-2.79; ,P, = .59). Conclusions: A pharmacist-led telephone outreach program conducted after ED discharge was not associated with a change in health care utilization.