Personalizing renal replacement therapy initiation in the intensive care unit: a reinforcement learning-based strategy with external validation on the AKIKI randomized controlled trials.

OBJECTIVE: The timely initiation of renal replacement therapy (RRT) for acute kidney injury (AKI) requires sequential decision-making tailored to individuals' evolving characteristics. To learn and validate optimal strategies for RRT initiation, we used reinforcement learning on clinical data from routine care and randomized controlled trials. MATERIALS AND METHODS: We used the MIMIC-III database for development and AKIKI trials for validation. Participants were adult ICU patients with severe AKI receiving mechanical ventilation or catecholamine infusion. We used a doubly robust estimator to learn when to start RRT after the occurrence of severe AKI for three days in a row. We developed a "crude strategy" maximizing the population-level hospital-free days at day 60 (HFD60) and a "stringent strategy" recommending RRT when there is significant evidence of benefit for an individual. For validation, we evaluated the causal effects of implementing our learned strategies versus following current best practices on HFD60. RESULTS: We included 3748 patients in the development set and 1068 in the validation set. Through external validation, the crude and stringent strategies yielded an average difference of 13.7 [95% CI -5.3 to 35.7] and 14.9 [95% CI -3.2 to 39.2] HFD60, respectively, compared to current best practices. The stringent strategy led to initiating RRT within 3 days in 14% of patients versus 38% under best practices. DISCUSSION: Implementing our strategies could improve the average number of days that ICU patients spend alive and outside the hospital while sparing RRT for many. CONCLUSION: We developed and validated a practical and interpretable dynamic decision support system for RRT initiation in the ICU.

Patients' Perspectives on How to Improve Endometriosis Care: A Large Qualitative Study Within the ComPaRe-Endometriosis e-Cohort.

Background: Endometriosis is a chronic gynecological condition that affects about 10% of women of reproductive age. Despite its prevalence, diagnosis is often delayed, misdiagnosis is common, and treatment options are poor. This study aimed at capturing ideas to improve endometriosis care from the patients' perspectives. Materials and Methods: We analyzed cross-sectional data from 1,000 adult patients in ComPaRe-Endometriosis (a French prospective e-cohort focused on endometriosis) who answered to the open-ended question: "If you had a magic wand, what would you change about your health care?". The free-text responses were analyzed by qualitative thematic analysis using an inductive approach. Results: Patients had a mean age of 34.1 years (standard deviation = 8.1); 56% and 42% had stage IV disease or deep endometriosis, respectively. They elicited 2,487 ideas to improve the management of endometriosis, which were categorized into 61 areas of improvement, further grouped into 14 themes. The top five areas of improvement were mentioned by >10% of the patients and were to (1) train caregivers to develop their knowledge on the disease, (2) provide better management of daily pain and pain attacks, (3) take patient-reported symptoms seriously, (4) standardize diagnostic processes to improve early detection, and (5) have caregivers listen more to the patients. Conclusions: We identified 61 areas for improvement in endometriosis care. These results reflect patients' expectations in terms of management of their disease and will be useful to design a better global care for endometriosis from the patients' perspectives.

Priorities to improve the care for chronic conditions and multimorbidity: a survey of patients and stakeholders nested within the ComPaRe e-cohort.

OBJECTIVE: To set priorities to improve consultations, care structures and the healthcare system from the perspective of patients with chronic conditions, care professionals, hospital administrators and health policy makers. METHODS: Adult patients with chronic conditions recruited from the ComPaRe e-cohort in France ranked their 15 most important areas of improvement among 147 previously defined by patients. Priorities at a population level were obtained by using logit models for sets of ranked items in a data set calibrated to represent the French population of patients with chronic conditions. Care professionals, hospital managers and health policy makers rated the complexity involved in improving the areas identified. We calculated the number of patients who considered as a priority at least one of the areas considered easy to implement. RESULTS: Between September 2018 and May 2019, 3002 patients (84% women, 47% with multimorbidity) and 149 professionals (including 50 care professionals, 79 hospital directors, 11 health policy decision makers) were recruited. Patients' top priorities were (1) Transforming care to be holistic and personalised, at a consultation level; (2) Smoothing patients' journey in the care system, increasing their knowledge of their own health and improving care coordination, at a care structure level (3) Training clinicians in better interpersonal skills and knowledge of specific conditions/treatments, reducing stigma and making care more affordable, at a healthcare system level. In total, 48%, 71% and 57% patients ranked in their top priorities one area considered easy to improve by professionals at consultation, care structure and health system levels, respectively. CONCLUSION: This is the first comprehensive map of patients' priorities to improve the management of chronic conditions. Implementing simple actions could benefit a large number of patients.