Machine Learning and Health Science Research: Tutorial.

Machine learning (ML) has seen impressive growth in health science research due to its capacity for handling complex data to perform a range of tasks, including unsupervised learning, supervised learning, and reinforcement learning. To aid health science researchers in understanding the strengths and limitations of ML and to facilitate its integration into their studies, we present here a guideline for integrating ML into an analysis through a structured framework, covering steps from framing a research question to study design and analysis techniques for specialized data types.

The Unseen Hand: AI-Based Prescribing Decision Support Tools and the Evaluation of Drug Safety and Effectiveness.

The use of artificial intelligence (AI)-based tools to guide prescribing decisions is full of promise and may enhance patient outcomes. These tools can perform actions such as choosing the 'safest' medication, choosing between competing medications, promoting de-prescribing or even predicting non-adherence. These tools can exist in a variety of formats; for example, they may be directly integrated into electronic medical records or they may exist in a stand-alone website accessible by a web browser. One potential impact of these tools is that they could manipulate our understanding of the benefit-risk of medicines in the real world. Currently, the benefit risk of approved medications is assessed according to carefully planned agreements covering spontaneous reporting systems and planned surveillance studies. But AI-based tools may limit or even block prescription to high-risk patients or prevent off-label use. The uptake and temporal availability of these tools may be uneven across healthcare systems and geographies, creating artefacts in data that are difficult to account for. It is also hard to estimate the 'true impact' that a tool had on a prescribing decision. International borders may also be highly porous to these tools, especially in cases where tools are available over the web. These tools already exist, and their use is likely to increase in the coming years. How they can be accounted for in benefit-risk decisions is yet to be seen.

Comparative Effect of Loop Diuretic Prescription on Mortality and Heart Failure Readmission.

Loop diuretics are a standard pharmacologic therapy in heart failure (HF) management. Although furosemide is most frequently used, torsemide and bumetanide are increasingly prescribed in clinical practice, possibly because of superior bioavailability. Few real-world comparative effectiveness studies have examined outcomes across all 3 loop diuretics. The study goal was to compare the effects of loop diuretic prescribing at HF hospitalization discharge on mortality and HF readmission. We identified patients in Medicare claims data initiating furosemide, torsemide, or bumetanide after an index HF hospitalization from 2007 to 2017. We estimated 6-month risks of all-cause mortality and a composite outcome (HF readmission or all-cause mortality) using inverse probability of treatment weighting to adjust for relevant confounders. We identified 62,632 furosemide, 1,720 torsemide, and 2,389 bumetanide initiators. The 6-month adjusted all-cause mortality risk was lowest for torsemide (13.2%), followed by furosemide (14.5%) and bumetanide (15.6%). The 6-month composite outcome risk was 21.4% for torsemide, 24.7% for furosemide, and 24.9% for bumetanide. Compared with furosemide, the 6-month all-cause mortality risk was 1.3% (95% confidence interval [CI]: -3.7, 1.0) lower for torsemide and 1.0% (95% CI: -1.2, 3.2) higher for bumetanide, and the 6-month composite outcome risk was 3.3% (95% CI: -6.3, -0.3) lower for torsemide and 0.2% (95% CI: -2.5, 2.9) higher for bumetanide. In conclusion, the findings suggested that the first prescribed loop diuretic following HF hospitalization is associated with clinically important differences in morbidity in older patients receiving torsemide, bumetanide, or furosemide. These differences were consistent for the effect of all-cause mortality alone, but were not statistically significant.

Beyond Two Cultures: Cultural Infrastructure for Data-driven Decision Support.

In the twenty years since Dr. Leo Breiman's incendiary paper Statistical Modeling: The Two Cultures was first published, algorithmic modeling techniques have gone from controversial to commonplace in the statistical community. While the widespread adoption of these methods as part of the contemporary statistician's toolkit is a testament to Dr. Breiman's vision, the number of high-profile failures of algorithmic models suggests that Dr. Breiman's final remark that "the emphasis needs to be on the problem and the data" has been less widely heeded. In the spirit of Dr. Breiman, we detail an emerging research community in statistics - data-driven decision support. We assert that to realize the full potential of decision support, broadly and in the context of precision health, will require a culture of social awareness and accountability, in addition to ongoing attention towards complex technical challenges.

Access to HIV prevention services in East African cross-border areas: a 2016-2017 cross-sectional bio-behavioural study.

INTRODUCTION: East African cross-border areas are visited by mobile and vulnerable populations, such as men, female sex workers, men who have sex with men, truck drivers, fisher folks and young women. These groups may not benefit from traditional HIV prevention interventions available at the health facilities where they live, but may benefit from services offered at public venues identified as places where people meet new sexual partners (e.g. bars, nightclubs, transportation hubs and guest houses). The goal of this analysis was to estimate availability, access and uptake of prevention services by populations who visit these venues. METHODS: We collected cross-sectional data using the Priorities for Local AIDS Control Efforts sampling method at cross-border locations near or along the land and lake borders of Kenya, Rwanda, Tanzania and Uganda from June 2016-February 2017. This bio-behavioural survey captured information from a probability sample of 11,428 individuals at 833 venues across all areas. Data were weighted using survey sampling weights and analysed using methods to account for the complex sampling design. RESULTS: Among the 85.6% of persons who had access to condoms, 60.5% did not use a condom at their last anal or vaginal sexual encounter. Venues visited by high percentages of persons living with HIV were not more likely than other venues to offer condoms. In 12 of the 22 cross-border areas, male or female condoms were available at less than 33% of the venues visited by persons having difficulty accessing condoms. In 17 of the 22 cross-border areas, education outreach visits in the preceding six months occurred at less than 50% of the venues where participants had low effective use of condoms. CONCLUSIONS: Individuals visiting venues in cross-border areas report poor access to and low effective use of condoms and other prevention services. Availability of HIV prevention services differed by venue and population type and cross-border area, suggesting opportunities for more granular targeting of HIV prevention interventions and transnational coordination of HIV programming.