Rethinking PICO in the Machine Learning Era: ML-PICO

Background  Machine learning (ML) has captured the attention of many clinicians who may not have formal training in this area but are otherwise increasingly exposed to ML literature that may be relevant to their clinical specialties. ML papers that follow an outcomes-based research format can be assessed using clinical research appraisal frameworks such as PICO (Population, Intervention, Comparison, Outcome). However, the PICO frameworks strain when applied to ML papers that create new ML models, which are akin to diagnostic tests. There is a need for a new framework to help assess such papers. Objective  We propose a new framework to help clinicians systematically read and evaluate medical ML papers whose aim is to create a new ML model: ML-PICO (Machine Learning, Population, Identification, Crosscheck, Outcomes). We describe how the ML-PICO framework can be applied toward appraising literature describing ML models for health care. Conclusion  The relevance of ML to practitioners of clinical medicine is steadily increasing with a growing body of literature. Therefore, it is increasingly important for clinicians to be familiar with how to assess and best utilize these tools. In this paper we have described a practical framework on how to read ML papers that create a new ML model (or diagnostic test): ML-PICO. We hope that this can be used by clinicians to better evaluate the quality and utility of ML papers.     

Attending Physicians' Perspectives of Resident Academic Half Day

Phenomenon: Academic health centers face significant challenges trying to improve medical education while meeting patient care needs. In response to problems with traditional forms of didactic education, many residency programs have transitioned to Academic Half Day (AHD), a curricular model in which learning is condensed into half-day blocks. In this model, trainees have protected educational time free from clinical responsibilities. However, an understanding of the impact on attending physicians and patient care when residents depart clinical sites for learning activities has not been well described. We sought to explore attending physicians’ perspectives when residents depart clinical sites to attend AHD. Approach: We performed a qualitative study with a grounded theory approach using individual semistructured interviews (December 2016–April 2017) of attending physicians who worked at inpatient and emergency department clinical sites from which residents departed to attend AHD. We used the constant comparative method, generating codes using an iterative approach and continuing sampling until saturation was reached. Major themes were identified and disagreements were resolved by consensus. Findings: Fifteen attending physicians from 6 clinical services were interviewed. Data analysis yielded 5 themes: emotional strain of workload, technology and systems challenges, patient safety and care concerns, disrupted resident learning, and the challenge to optimize resident education. Attending physicians, already working on busy services, felt frustrated and perceived having an increased workload when residents departed for AHD. They were concerned about safely entering orders in the electronic health record, impeded patient workflow, and further disruption of resident schedules already disrupted by duty hour restrictions and continuity clinic. Attending physicians described the importance of experiential learning from caring for patients and from structured didactic learning; however, the optimal balance was uncertain. Insights: We found that attending physicians experienced significant emotional strain, faced technological challenges, and were concerned about impeded workflow and patient safety when residents departed clinical sites for AHD. This is likely to be true whenever residents are pulled out of the clinical setting for any reason. Educators need to partner with hospital administrators to provide appropriate support for attending physicians when residents leave clinical sites, evaluate the effectiveness of different educational models, and determine how structured learning activities fit into the overall curriculum.     

Participatory research in Northwest Atlanta's Proctor Creek Watershed: Using photovoice to explore environmental health risks at the water's edge

In this study we used a participatory research method, photovoice, to explore community perceptions about environmental health risks, community assets, and strengths in and around an urban, degraded watershed in Northwest Atlanta, Georgia. This watershed, formed by Proctor Creek, is a focal point for redevelopment and infrastructure investments for years to come. Using a community-based participatory research approach, 10 Proctor Creek residents (watershed researchers), and a university partner, engaged in data collection; participatory data analysis; internal discussions; translation of research findings into watershed restoration, community revitalization, remedial action, and policy solutions; and dissemination of results to fellow watershed residents, stakeholders, and decision makers. We present a conceptual model linking the watershed researchers' understanding of urban policies and practice in the Proctor Creek Watershed to environmental, neighborhood and housing conditions and their influence on health outcomes and quality of life. Engaging community members in defining their own community environmental health challenges and assets yielded the following primary themes: 1) threats to the natural environment, 2) built environment stressors that influence health, 3) blight and divestment of public resources, and 4) hope for the future. Residents’ vision for the future of the watershed - a restored creek, revitalized neighborhoods, and restored people - is fueled by a strong connection to history, memory, and sense of place. We demonstrate the value of local knowledge in identifying previously unaddressed environmental health risks in the Proctor Creek Watershed as well as solutions to reduce or eliminate them.