Joanna Abraham

Objectives: To investigate the longitudinal effects of an ambient artificial intelligence (AI) scribe use in primary care on outcomes related to documentation time, note writing patterns, and clinician financial productivity. Materials and Methods: We conducted a quasi-experimental study with an interrupted time-series design to ascertain the effect of AI scribe use on note writing time, time outside of scheduled hours (TOSH), pajama time (5:30PM-7:00aM), note length, note closure <24 h, and work relative value units (wRVU). For each primary care clinician (physician, advanced practice provider), date of initial access was day 0; each measure was calculated for 150 days prior (pre-period) and after (post-period) day zero. Results: Two hundred and twenty clinicians (mean age = 43.7, 70.9% females; 56.4% physicians) from 36 clinics, conducting 314 845 patient encounters were included. Compared to the pre-period, there was 7% decrease in note writing time at day 0 (incidence rate ratio [IRR] 0.93, 95% credible interval [CI] [0.89, 0.98]), increasing to 16% decrease by day 150 (IRR 0.84, 95% CI [0.82, 0.87]). There was no change in pajama time, TOSH, or wRVU at day 0; however, at day 150, there was a 19% decrease in pajama time (0.81, 95% CI [0.73, 0.90]), 13% decrease in TOSH (0.87, 95% CI [0.77, 0.98]), and a 2% increase in wRVU (1.02, 95% CI [1.01, 1.04]). At day 0, there was a 5% increase in note closures (1.05, 95% CI [1.02, 1.08]), with no difference to pre-period levels by day 150. There were no changes in note length at day 0 or day 150. Discussion and Conclusion: There were persistent changes in AI scribe use measures over time as clinicians situated these tools within their workflows.

Background and Objective: Cardiac surgeries are associated with a range of serious postoperative complications, and current predictive models often rely on static summaries of preoperative and intraoperative data compiled after the case. Such development limits their utility in real-time clinical settings. To address this challenge, we developed TE-Transformer, an end-to-end deep learning framework that continuously ingests heterogeneous electronic health record (EHR) data—including preoperative variables, high-frequency vital signs, medication administration records, and brief surgical notes texts—to provide real-time, multitask predictions of multiple postoperative complications. Methods: A retrospective cohort of 6416 patients undergoing cardiac surgery was used to develop and evaluate TE-Transformer. The architecture reduced reliance on manually engineered temporal summary features by encoding irregular intraoperative observations as event-level representations. Continuous risk estimates were updated at 30-minute intervals, enabling a dynamic perspective on patient status. Performance was benchmarked against traditional machine learning models (eg, XGBoost, Random Forest) and deep learning comparators (LSTM, GRU, CNN) frequently used in sequential modelling. Results: We analyzed 6416 cardiac surgeries (61% male; median [IQR] age 62 [53-70]). Across classification tasks, TE-Transformer achieved AUROC (area under the receiver operating characteristic curve)/AUPRC (area under the precision-recall curve) of 0.8461/0.3994 (mortality), 0.8837/0.8033 (delirium), and 0.7956/0.6666 (acute kidney injury). For continuous endpoints, mean absolute error (MAE) was 0.0633 (low SBP [systolic blood pressure] time), 0.0920 (N glucose high), 0.0020 (N glucose low), and 0.9781 (postoperative ventilation duration). Ablation study showed the full multimodal model improved performance over the preoperative-only baseline, most notably for 30-day mortality AUROC (+0.049), postoperative delirium AUPRC (+0.059), and postoperative ventilation MAE (−0.082), with intraoperative flowsheet events contributing the largest single increment. Conclusion: TE-Transformer used routinely collected perioperative data to generate continuously updated risk estimates for multiple postoperative complications and achieved better predictive performance than conventional models. These findings support further evaluation of real-time, multimodal prediction frameworks for intraoperative decision support. This approach highlights the potential of transformer-based, multitask prediction as a foundation for future intraoperative decision support in cardiac surgery.

Background: Health-related technology use among nondigitally native adults is becoming widespread. Nevertheless, digital health interventions are typically not designed with the unique usability needs and preferences of this population in mind. Furthermore, most of these middle-aged and older adults are managing multiple medical conditions, which can also impact their preferences related to digital health interventions. A prime opportunity to address multimorbidity in nondigital natives using a digital health intervention is the intersection between mental health and chronic pain. Objective: The goal of this study was to use established frameworks and end-user feedback to identify actionable, usability-related features that can be incorporated into existing digital health interventions and that are preferred among nondigitally native adults. We aimed to identify general usability adaptations, as well as just-in-time adaptive interventions (JITAIs). Methods: We conducted a qualitative usability study of a convenience sample using a hybrid inductive-deductive content analysis to evaluate an existing mental health app (Wysa for Chronic Pain). Participants were 45 years or older; reported at least moderate symptoms of depression or anxiety (Patient Health Questionnaire -9 ≥10 or Generalized Anxiety Disorder -7 score ≥10); and endorsed having pain at least most days in the past 3 months. The Framework for Reporting Adaptations and Modifications to Evidence-based Implementation Strategies was used to identify potential usability-related adaptations for the target population. Development of usability-related JITAIs was guided by Nahum-Shani’s pragmatic framework for JITAI development and the Behavioral Integration Technology model. Results: Forty-two participants completed usability testing (mean age 57, SD 8 years; women n=32, 76%). Participants identified numerous opportunities to optimize their user experience, primarily by ensuring the app clearly describes how all its features are intended to be used and by minimizing navigation burden. Specifically, participants requested clear, step-by-step orientation and navigation instructions, rather than a brief onboarding experience that relies on user-led exploration and familiarity with conventional app symbols. Participants also recommended a customized experience based on their unique usage patterns. The most common JITAI opportunity identified was to strategically reduce user notifications in order to reduce the risk of notification fatigue and subsequent complete disengagement with the app. Conclusions: Identifying actionable opportunities to improve usability and prompt engagement with “the right tool at the right time for the right person” holds promise to improve the effectiveness of digital health interventions across the age span. The usability-related refinement opportunities identified in this study are generalizable to other digital health interventions that are relevant to older users who are likely to be managing multimorbidity, are not digital natives, and are more likely than younger users to have physical or cognitive challenges. The integrated framework-based process described in this article can also serve as a model for optimizing other existing digital health interventions.

Background: On a population level, mental health apps are accessible and effective. However, nondigitally native adults with chronic pain are a large and growing population who have been neglected during the development process of these interventions. Although technology use is rapidly growing among this population, their engagement with mobile health–related apps is lagging because usability is often not optimized for their needs and preferences. Objective: This study aimed to identify design preferences and determinants of engagement with mental health apps by nondigitally native adults who have chronic pain and coexisting symptoms of depression or anxiety. Methods: In this qualitative study, participants completed a semistructured interview regarding their experience with, and perceptions of, mobile devices, apps, and digital health interventions. Participants were 45 years or older; scored ≥10 on the 9-item Patient Health Questionnaire, 7-item Generalized Anxiety Disorder, or both; endorsed pain on most days or every day in the past 3 months; and were living in the United States. The interview guide was informed by the Consolidated Framework for Implementation Research and the Behavioral Intervention Technology model. Codes were organized into themes. Recruitment continued until thematic saturation was achieved. Results: A total of 42 participants were interviewed (mean age 57, SD 8 years; n=32, 76% women). Participants strongly preferred apps that are free, describe strong privacy policies, and add functional value to their lives. They were more motivated by “real-life” goal achievement and tangible health improvements than by gamification within an app, and many were wary of allowing apps to passively collect certain types of data, especially to make inferences about their mental health. Most participants were unaware of apps designed to address chronic pain, but they were interested in the concept, particularly to help track their mood and pain and then identify associations between their symptoms, app activity, and other life events. Despite their daily use of apps, many participants described frequent challenges related to app navigation. While on-demand access to app tools was preferred, most participants appreciated the potential value of occasional push notifications if their timing was thoughtful and customizable. Participants cautioned against an overly cheerful, “infantile,” or informal tone for an app that addresses serious issues such as mental health and chronic pain. Conclusions: Mental health apps for nondigitally native adults should highlight tangible health improvements that can be achieved from app engagement (more so than gamification), potentially using a multidomain tracking feature, if appropriate. This population is available to receive just-in-time adaptive interventions, but the frequency and timing should be thoughtful, customizable, and not wholly reliant on passively collected personal data. Health-related apps designed to address conditions that are more common with increasing age should account for these preferences.

Objectives: To explore patient experiences with a perioperative mental health intervention (PMHI), understand its perceived impact on emotional health, and identify factors affecting engagement and feasibility. Background: Depression and anxiety can worsen surgical outcomes in older adults, highlighting the need for perioperative emotional health support. We evaluated a PMHI across 3 linked randomized controlled trials in cardiac, oncologic, and orthopedic surgery. Here, we report qualitative findings highlighting patients’ experiences with the PMHI across trials. Methods: The study was conducted across 3 hospitals in a Midwestern healthcare network. 112 semi-structured interviews were conducted with older patients undergoing surgeries (31 cardiac, 39 oncologic, and 42 orthopedic), complemented by 5 periodic reflection focus groups with 9 interventionists (4 social workers/counselors and 5 pharmacist team members). An inductive thematic approach was used for analysis. Results: Patients and interventionists perceived the PMHI positively, identifying numerous beneficial factors such as compassionate care, strong rapport, and thorough medication education. However, patients noted nonbeneficial factors limiting engagement with the PMHI, including fatigue and competing personal priorities. Patients with oncologic or complex cardiac procedures reported the greatest benefit from the PMHI, compared with slightly lower perceived usefulness among orthopedic and simpler cardiac patients. Informed by these findings, we present a conceptual model illustrating beneficial and nonbeneficial factors influencing patient perceptions of the PMHI. Conclusions: Patients’ positive perceptions of the PMHI and its impact on alleviating depression and anxiety symptoms underscore the value of integrating psychological and pharmacological support into perioperative care for older adults.

Introduction: Mixed methods are used to holistically understand the “what”, “how”, and “why” questions within a single study by integrating quantitative and qualitative methods. Although this approach has demonstrated value in other disciplines, their use in applied clinical informatics research remains largely unexplored. This methodological review characterized the types of informatics intervention studies that used mixed methods designs and the specific methods of evaluation and reporting. Materials and Methods: On December 13, 2024, we searched Embase, MEDLINE, PubMed, CINAHL and PsycINFO for studies that reportedly used a mixed methods design to study the development or evaluation of clinical informatics interventions. We developed themes in topics studied, designs used, and reporting issues. Quality assessment was conducted using the Mixed Methods Appraisal Tool. Results: We included 54 studies. Despite being described as mixed methods, 14 studies (25.9%) lacked the integration of quantitative and qualitative data. Of the remaining studies, convergent study designs were commonly used. Mixed methods were predominantly used during pre-implementation and post-implementation evaluations. Quality issues included non-representative samples and non-reporting of qualitative approaches and paradigms. Discussion and Conclusions: The use of mixed methods to study clinical informatics development and implementation is uncommon. This review confirmed that reporting and design problems found in other disciplines extend to informatics. Calls to action include a need to disseminate mixed methods guidance within the informatics community and offer mixed methods training in graduate programs. We also synthesized an initial reporting checklist from the literature. Our findings offer baseline levels to assess educational efforts.

Importance: Clinician adoption and adaptation of new tools evolve over time. Prior studies of ambient Artificial intelligence (AI) scribes have primarily relied on single time-point measurements (e.g., pre-post), potentially obfuscating their true impact on outcomes. Objective: To investigate longitudinal effects of an AI scribe tool on patient encounter-level outcomes. Design: Case series across 48 weeks (24 pre, 24 post) per clinician. Setting: Primary care clinical encounters occurring between 01/05/24 and 10/31/25. Participants: Primary care clinicians (attending physicians and advanced practice providers). Exposure: Ambient AI scribe introduction to clinical workflow, indexed to study day zero, per clinician. Main outcomes and measures: Encounter-level measurements of documentation time (note writing time, time outside of scheduled hours (TOSH), pajama time), note writing patterns (note length, note closure <24h) and clinician’s billed work Relative Value Units (wRVU) with a focus on changes from pre-period outcomes at Day 0 and 150. Results: 220 primary care clinicians (Mean age=43.7, 70.9% females; 56.4% physicians) from 36 clinics, conducting 314,845 patient encounters were included. All outcomes evolved from day zero to day 150 and are compared back to pre-period levels. There was evidence of an immediate 7% decrease on average in note writing time at day zero (Incidence Rate Ratio, IRR 0.93, 95%CI [0.89, 0.96]), intensifying to a 15% decrease by day 150 (IRR 0.85, 95%CI [0.83, 0.87]). There was no evidence of a change in pajama time or TOSH at day zero; however, at day 150, there was evidence of a 18% decrease in pajama time (0.82, 95%CI [0.73, 0.91]) and a 13% decrease in TOSH (0.87, 95%CI [0.77, 0.99]). At day zero, there was evidence of a 5% increase (1.05, 95%CI [1.00, 1.10]) in note length and 31% increase in note closures (1.31, 95%CI [1.13, 1.53]), with both slowly attenuating to pre-period levels by day 150. Although there was no evidence of changes in wRVU at day zero, there was a 2% increase total wRVU at day 150 (1.02, 95%CI [1.01, 1.03]). Conclusions and relevance: Longitudinal changes were gradual, but persistent, underscoring the gradual adaptation of AI scribes, as clinicians situated these tools within their workflows. Key Points: Question: How do the patterns of use of an ambient Artificial Intelligence (AI) scribe evolve over time? Findings: In this longitudinal, quasi-experimental study on clinician use of an ambient AI scribe, documentation time, note writing patterns and financial productivity evolved over a 150-day period. Compared to the pre-period, note writing time savings increased from 7% (day zero) to 15% (day 150); changes in all other considered outcomes including time outside of scheduled hours, pajama time, note length, note closure <24h, billed work Relative Value Units evolved over the 150-day period. Meaning: Clinician use of ambient AI scribes showed persistent changes in patterns of use over a 150-day period, highlighting a gradual adaptation process and the need for longitudinal assessment.

Background: Chronic obstructive pulmonary disease (COPD) affects approximately 16 million Americans and often results in avoidable readmissions due, in part, to medication errors and lack of education. Telehealth interventions can support medication reconciliation and inhaler education following hospital discharge for patients with COPD. Objective: This study aimed to design and prototype TELE-TOC (Telehealth Education: Leveraging Electronic Transitions of Care), a post-discharge, in-home, televisit intervention, and to map its workflow to ensure integration into the routine discharge care transition process for patients with COPD. Methods: A user-centered design approach across 3 phases was followed to develop and prototype TELE-TOC. Participants included adult patients hospitalized for COPD exacerbations, their caregivers, clinicians involved in COPD care, and organizational leaders. Data collection methods included semi-structured interviews, system usability scale surveys, and cognitive walkthroughs of the TELE-TOC prototype to assess participants’ perceptions on usability and feasibility of TELE-TOC implementation as part of routine COPD discharge care transitions. Qualitative data were analyzed using inductive thematic analysis and an inductive-deductive approach guided by the Agency for Healthcare Research and Quality–endorsed Care Transitions Framework. Quantitative data were summarized using basic descriptive statistics. Results: Participants included 18 patients, 18 clinicians, 8 organizational leaders, and 2 caregivers. Phase 1 identified 3 interdependent stages of COPD hospital-to-home discharge: inpatient pre-discharge, at-home post-discharge, and outpatient clinic visit post-discharge. Key facilitators of discharge care transitions included the hospital’s “meds-to-beds” program and high patient health literacy, while barriers to discharge included poor timing of education and conflicting patient priorities. Phase 2 delineated the core televisit components (eg, dedicated clinician, medication reconciliation, inhaler use, and self-management education) and flexible components (eg, reminder system and session frequency). Potential implementation enablers included multiple techniques for clinicians to access and support patient education and backup communication strategies in the event of technical issues. Potential implementation barriers included insufficient patient technology access and limited technology and health literacy, as well as limited clinician bandwidth for thorough COPD education and medication reconciliation. Phase 3 TELE-TOC prototype walkthroughs demonstrated a positive patient experience (average system usability scale score of 97.5/100), attributed to the benefits of videoconferencing technology for hands-on teaching and the use of the virtual teach-back method. Identified barriers included varying levels of patient technology literacy, insufficient inhaler education, limited patient understanding of medication lists, and clinician uncertainty around TELE-TOC documentation. Suggestions for mitigating these barriers included patient training for TELE-TOC sessions, amendments to pharmacists’ “visit note,” and enhanced patient preparation for medication reconciliation. Conclusions: Using a co-design approach, we integrated multiple perspectives to develop and optimize TELE-TOC, a patient-centered televisit intervention aimed at supporting discharge care transitions to improve continuity of care and outcomes for patients with COPD. Future research will evaluate the impact of TELE-TOC on readmissions from acute exacerbations.

BACKGROUND On a population level, mental health apps are accessible and effective. However, nondigitally native adults with chronic pain are a large and growing population who have been neglected during the development process of these interventions. Although technology use is rapidly growing among this population, their engagement with mobile health–related apps is lagging because usability is often not optimized for their needs and preferences. OBJECTIVE This study aimed to identify design preferences and determinants of engagement with mental health apps by nondigitally native adults who have chronic pain and coexisting symptoms of depression or anxiety. METHODS In this qualitative study, participants completed a semistructured interview regarding their experience with, and perceptions of, mobile devices, apps, and digital health interventions. Participants were 45 years or older; scored ≥10 on the 9-item Patient Health Questionnaire, 7-item Generalized Anxiety Disorder, or both; endorsed pain on most days or every day in the past 3 months; and were living in the United States. The interview guide was informed by the Consolidated Framework for Implementation Research and the Behavioral Intervention Technology model. Codes were organized into themes. Recruitment continued until thematic saturation was achieved. RESULTS A total of 42 participants were interviewed (mean age 57, SD 8 years; n=32, 76% women). Participants strongly preferred apps that are free, describe strong privacy policies, and add functional value to their lives. They were more motivated by “real-life” goal achievement and tangible health improvements than by gamification within an app, and many were wary of allowing apps to passively collect certain types of data, especially to make inferences about their mental health. Most participants were unaware of apps designed to address chronic pain, but they were interested in the concept, particularly to help track their mood and pain and then identify associations between their symptoms, app activity, and other life events. Despite their daily use of apps, many participants described frequent challenges related to app navigation. While on-demand access to app tools was preferred, most participants appreciated the potential value of occasional push notifications if their timing was thoughtful and customizable. Participants cautioned against an overly cheerful, “infantile,” or informal tone for an app that addresses serious issues such as mental health and chronic pain. CONCLUSIONS Mental health apps for nondigitally native adults should highlight tangible health improvements that can be achieved from app engagement (more so than gamification), potentially using a multidomain tracking feature, if appropriate. This population is available to receive just-in-time adaptive interventions, but the frequency and timing should be thoughtful, customizable, and not wholly reliant on passively collected personal data. Health-related apps designed to address conditions that are more common with increasing age should account for these preferences.