Clinical Needs Assessment of a Machine Learning-Based Asthma Management Tool: User-Centered Design Approach.

BACKGROUND: Personalized asthma management depends on a clinician's ability to efficiently review patient's data and make timely clinical decisions. Unfortunately, efficient and effective review of these data is impeded by the varied format, location, and workflow of data acquisition, storage, and processing in the electronic health record. While machine learning (ML) and clinical decision support tools are well-positioned as potential solutions, the translation of such frameworks requires that barriers to implementation be addressed in the formative research stages. OBJECTIVE: We aimed to use a structured user-centered design approach (double-diamond design framework) to (1) qualitatively explore clinicians' experience with the current asthma management system, (2) identify user requirements to improve algorithm explainability and Asthma Guidance and Prediction System prototype, and (3) identify potential barriers to ML-based clinical decision support system use. METHODS: At the "discovery" phase, we first shadowed to understand the practice context. Then, semistructured interviews were conducted digitally with 14 clinicians who encountered pediatric asthma patients at 2 outpatient facilities. Participants were asked about their current difficulties in gathering information for patients with pediatric asthma, their expectations of ideal workflows and tools, and suggestions on user-centered interfaces and features. At the "define" phase, a synthesis analysis was conducted to converge key results from interviewees' insights into themes, eventually forming critical "how might we" research questions to guide model development and implementation. RESULTS: We identified user requirements and potential barriers associated with three overarching themes: (1) usability and workflow aspects of the ML system, (2) user expectations and algorithm explainability, and (3) barriers to implementation in context. Even though the responsibilities and workflows vary among different roles, the core asthma-related information and functions they requested were highly cohesive, which allows for a shared information view of the tool. Clinicians hope to perceive the usability of the model with the ability to note patients' high risks and take proactive actions to manage asthma efficiently and effectively. For optimal ML algorithm explainability, requirements included documentation to support the validity of algorithm development and output logic, and a request for increased transparency to build trust and validate how the algorithm arrived at the decision. Acceptability, adoption, and sustainability of the asthma management tool are implementation outcomes that are reliant on the proper design and training as suggested by participants. CONCLUSIONS: As part of our comprehensive informatics-based process centered on clinical usability, we approach the problem using a theoretical framework grounded in user experience research leveraging semistructured interviews. Our focus on meeting the needs of the practice with ML technology is emphasized by a user-centered approach to clinician engagement through upstream technology design.

Stakeholder perceptions of using "opt-out" for tobacco use treatment in a cancer care setting: a qualitative evaluation of patients, providers, and desk staff.

BACKGROUND: Continued tobacco use in cancer patients increases the risk of cancer treatment failure and decreases survival. However, currently, most cancer patients do not receive evidence-based tobacco treatment. A recently proposed "opt-out" approach would automatically refer all cancer patients who use tobacco to tobacco treatment, but its acceptability to cancer patients and providers is unknown. We aimed to understand stakeholder beliefs, concerns, and receptivity to using the "opt-out" approach for tobacco treatment referrals in a cancer care setting. METHODS: Semi-structured interviews were conducted with oncology patients, providers, and desk staff. The sample size was determined when theoretical saturation was reached. Given the differences among participant roles, separate interview guides were developed. Transcripts were analyzed using standard coding techniques for qualitative data using the Consolidated Framework for Implementation Research (CFIR) codebook. Emergent codes were added to the codebook to account for themes not represented by a CFIR domain. Coded transcripts were then entered into the qualitative analysis software NVivo to generate code reports for CFIR domains and emergent codes for each stakeholder group. Data were presented by stakeholder group and subcategorized by CFIR domains and emergent codes when appropriate. RESULTS: A total of 21 providers, 19 patients, and 6 desk staff were interviewed. Overall acceptance of the "opt out" approach was high among all groups. Providers overwhelmingly approved of the approach as it requires little effort from them to operate and saves clinical time. Desk staff supported the opt-out system and believed there are clinical benefits to patients receiving information about tobacco treatment. Many patients expressed support for using an opt-out approach as many smokers need assistance but may not directly ask for it. Patients also thought that providers emphasizing the benefits of stopping tobacco use to cancer treatment and survival would be an important factor motivating them to attend treatment. CONCLUSIONS: While providers appreciated that the system required little effort on their part, patients clearly indicated that promotion of tobacco cessation treatment by their provider would be vital to enhance willingness to engage with treatment. Future implementation efforts of opt-out systems will require implementation strategies that promote provider engagement with their patients around smoking cessation while continuing to limit burden on providers.

A Technical Performance Study and Proposed Systematic and Comprehensive Evaluation of an ML-based CDS Solution for Pediatric Asthma.

Achieving optimal care for pediatric asthma patients depends on giving clinicians efficient access to pertinent patient information. Unfortunately, adherence to guidelines or best practices has shown to be challenging, as relevant information is often scattered throughout the patient record in both structured data and unstructured clinical notes. Furthermore, in the absence of supporting tools, the onus of consolidating this information generally falls upon the clinician. In this study, we propose a machine learning-based clinical decision support (CDS) system focused on pediatric asthma care to alleviate some of this burden. This framework aims to incorporate a machine learning model capable of predicting asthma exacerbation risk into the clinical workflow, emphasizing contextual data, supporting information, and model transparency and explainability. We show that this asthma exacerbation model is capable of predicting exacerbation with an 0.8 AUC-ROC. This model, paired with a comprehensive informatics-based process centered on clinical usability, emphasizes our focus on meeting the needs of the clinical practice with machine learning technology.

Docosahexaenoic acid decreased inflammatory gene expression, but not 18-kDa translocator protein binding, in rat pup brain after controlled cortical impact.

BACKGROUND: Traumatic brain injury is the leading cause of acquired neurologic disability in children. In our model of pediatric traumatic brain injury, controlled cortical impact (CCI) in rat pups, docosahexaenoic acid (DHA) improved lesion volume and cognitive testing as late as postinjury day (PID) 50. Docosahexaenoic acid decreased proinflammatory messenger RNA (mRNA) in microglia and macrophages at PIDs 3 and 7, but not 30. We hypothesized that DHA affected inflammatory markers differentially relative to impact proximity, early and persistently after CCI. METHODS: To provide a temporal snapshot of regional neuroinflammation, we measured 18-kDa translocator protein (TSPO) binding using whole brain autoradiography at PIDs 3, 7, 30, and 50. Guided by TSPO results, we measured mRNA levels in contused cortex and underlying hippocampus for genes associated with proinflammatory and inflammation-resolving states at PIDs 2 and 3. RESULTS: Controlled cortical impact increased TSPO binding at all time points, most markedly at PID 3 and in regions closest to impact, not blunted by DHA. Controlled cortical impact increased cortical and hippocampal mRNA proinflammatory markers, blunted by DHA at PID 2 in hippocampus. CONCLUSION: Controlled cortical impact increased TSPO binding in the immature brain in a persistent manner more intensely with more severe injury, not altered by DHA. Controlled cortical impact increased PIDs 2 and 3 mRNA levels of proinflammatory and inflammation-resolving genes. Docosahexaenoic acid decreased proinflammatory markers associated with inflammasome activation at PID 2. We speculate that DHA's salutary effects on long-term outcomes result from early effects on the inflammasome. Future studies will examine functional effects of DHA on microglia both early and late after CCI.

Design and Pilot Implementation of an Electronic Health Record-Based System to Automatically Refer Cancer Patients to Tobacco Use Treatment.

Continued tobacco use after cancer diagnosis is detrimental to treatment and survivorship. The current reach of evidence-based tobacco treatments in cancer patients is low. As a part of the National Cancer Institute Cancer Center Cessation Initiative, the Mayo Clinic Cancer Center designed an electronic health record (EHR, Epic©)-based process to automatically refer ambulatory oncology patients to tobacco use treatment, regardless of intent to cease tobacco use("opt out"). The referral and patient scheduling, accomplished through a best practice advisory (BPA) directed to staff who room patients, does not require a co-signature from clinicians. This process was piloted for a six-week period starting in July of 2019 at the Division of Medical Oncology, Mayo Clinic, Rochester, MN. All oncology patients who were tobacco users were referred for tobacco treatment by the rooming staff (n = 210). Of these, 150 (71%) had a tobacco treatment appointment scheduled, and 25 (17%) completed their appointment. We conclude that an EHR-based "opt-out" approach to refer patients to tobacco dependence treatment that does not require active involvement by clinicians is feasible within the oncology clinical practice. Further work is needed to increase the proportion of scheduled patients who attend their appointments.