Harnessing the Power of Generative AI for Clinical Summaries: Perspectives From Emergency Physicians.

STUDY OBJECTIVE: The workload of clinical documentation contributes to health care costs and professional burnout. The advent of generative artificial intelligence language models presents a promising solution. The perspective of clinicians may contribute to effective and responsible implementation of such tools. This study sought to evaluate 3 uses for generative artificial intelligence for clinical documentation in pediatric emergency medicine, measuring time savings, effort reduction, and physician attitudes and identifying potential risks and barriers. METHODS: This mixed-methods study was performed with 10 pediatric emergency medicine attending physicians from a single pediatric emergency department. Participants were asked to write a supervisory note for 4 clinical scenarios, with varying levels of complexity, twice without any assistance and twice with the assistance of ChatGPT Version 4.0. Participants evaluated 2 additional ChatGPT-generated clinical summaries: a structured handoff and a visit summary for a family written at an 8th grade reading level. Finally, a semistructured interview was performed to assess physicians' perspective on the use of ChatGPT in pediatric emergency medicine. Main outcomes and measures included between subjects' comparisons of the effort and time taken to complete the supervisory note with and without ChatGPT assistance. Effort was measured using a self-reported Likert scale of 0 to 10. Physicians' scoring of and attitude toward the ChatGPT-generated summaries were measured using a 0 to 10 Likert scale and open-ended questions. Summaries were scored for completeness, accuracy, efficiency, readability, and overall satisfaction. A thematic analysis was performed to analyze the content of the open-ended questions and to identify key themes. RESULTS: ChatGPT yielded a 40% reduction in time and a 33% decrease in effort for supervisory notes in intricate cases, with no discernible effect on simpler notes. ChatGPT-generated summaries for structured handoffs and family letters were highly rated, ranging from 7.0 to 9.0 out of 10, and most participants favored their inclusion in clinical practice. However, there were several critical reservations, out of which a set of general recommendations for applying ChatGPT to clinical summaries was formulated. CONCLUSION: Pediatric emergency medicine attendings in our study perceived that ChatGPT can deliver high-quality summaries while saving time and effort in many scenarios, but not all.

Factors associated with first-time and repeat blood donation: Adverse reactions and effects on donor behavior.

BACKGROUND: Blood centers have a dual mission to protect donors and patients; donor safety is paramount to maintaining an adequate blood supply. Elucidating donor factors associated with adverse reactions (AR) is critical to this mission. STUDY DESIGN/METHODS: A retrospective cohort analysis of whole blood donors from 2003 to 2020 was conducted at a single blood center in northern California. Adjusted odds ratios (AORs) with 95% CIs for ARs were estimated via multivariable logistic regression on demographics, donation history, and physical examination data. Where appropriate, Wilcoxon-Rank Sum and chi-squared tests were used to determine significance. RESULTS: First-time blood donors (FTD) exhibited a higher AR rate than repeat donors (4.4% vs. 1.9% p < .0001). When compared with FTDs without AR, FTDs with ARs (FT-AR) were less likely to return (30.0% vs. 47.3%, p < .0001), and, of those who returned, had a higher rate of reaction 20.2% versus 2.8% (p < .001). Factors found to be associated with FT-AR (younger age, increased heart rate, and higher diastolic blood pressure) still correlated positively with AR on return donation, but to a lower degree. FTD who potentially witnessed an AR had a lower return rate (44.6% vs. 47.3%, p = <.001) and donated fewer units (2.38 vs. 3.37, p < .001) when compared to FTD who did not witness an AR. CONCLUSION: The AR on FTD increases the AR likelihood of return donation. Longitudinal analysis shows that a time-based deferral policy targeted at FT-AR young donors can reduce the number of ARs while not dramatically impacting the blood supply.

Physician Risk Perception and Testing Behaviors for Children With Fever.

OBJECTIVES: Risk tolerance and risk perceptions may impact clinicians' decisions to obtain diagnostic tests. We sought to determine whether physician risk perception was associated with the decision to obtain blood or imaging tests among children who present to the emergency department with fever. METHODS: We conducted a retrospective, cross-sectional study in the Boston Children's Hospital emergency department. We included children aged 6 months to 18 years from May 1, 2014 to April 30, 2019, with fever. Our primary outcome was diagnostic testing: obtaining a blood and/or imaging test. We assessed risk perception using 3 scales: the Risk Tolerance Scale (RTS), Stress From Uncertainty Scale (SUS), and Malpractice Fear Scale (MFS). A z score was assigned to each physician for each scale. Mixed-effects logistic regression assessed the association between physician risk perception and blood or imaging testing. We also examined the relationship between each risk perception scale and several secondary outcomes: blood testing, urine testing, diagnostic imaging, specialist consultation, hospitalization, and revisit within 72 hours. RESULTS: The response rate was 55/56 (98%). We analyzed 12,527 encounters. Blood/imaging testing varied between physicians (median, 48%; interquartile range, 41%-53%; range, 30%-71%). Risk Tolerance Scale responses were not associated with blood/imaging testing (odds ratio [OR], 1.03 per SD of increased risk perception; 95% confidence interval [CI], 0.95-1.13). Stress From Uncertainty Scale responses were not associated with blood/imaging testing (OR, 1.04 per SD; 95% CI, 0.95-1.14). Malpractice Fear Scale responses were not associated with blood/imaging testing (OR, 1.00 per SD; 95% CI, 0.91-1.09). There was no significant association between RTS, MFS, or SUS and any secondary outcome, except that there was a weak association between SUS and specialist consultation (OR, 1.12; 95% CI, 1.00-1.24). CONCLUSIONS: Across 55 pediatric emergency physicians with variable testing practices, there was no association between risk perception and blood/imaging testing in febrile children.

Improving the prediction of streptococcal pharyngitis; time to move past exudate alone.

BACKGROUND: Palatal petechiae are predictive of Group A streptococcal (GAS) pharyngitis. We sought to (a) quantify the value of considering petechiae in addition to exudate, and (b) assess provider incorporation of petechiae's predictive nature for GAS into clinical decision making. METHODS: We conducted a cross-sectional study of patients 3-21 years with sore throat and GAS testing performed in a pediatric emergency department (ED) in 2016. Patients were excluded if immunosuppressed, nonverbal, medically complex, had chronic tonsillitis, or received antibiotics in the preceding week. As a proxy of provider incorporation of petechiae into clinical decision making we assessed how often petechiae were documented, compared with exudate. We performed univariate analysis using χ2 analysis for categorical data and Mann-Whitney U test for continuous data. RESULTS: 1574 patients met inclusion criteria. Median age 8 years [IQR 5, 13]; 54% female. 372 patients (24%) were GAS positive. Both palatal petechiae and tonsillar exudates were predictive of GAS [OR 8.5 (95% CI 5.2-13.9), and 1.9 (95% CI 1.4-2.6) respectively]. Examining petechiae or exudate vs. exudate alone increases OR from 1.9 to 2.9 (95% CI 2.2-3.8). Sensitivity improves (23% to 34%) with minimal change to specificity (87% to 85%). Among those with a normal or erythematous throat exam, petechiae were mentioned as a pertinent negative in 28%; absence of tonsillar exudate was mentioned in 78% (p = .02). CONCLUSIONS: Palatal petechiae are highly associated with GAS, yet rarely addressed in documentation. Incorporating palatal petechiae into common scoring systems could improve prediction and disseminate this knowledge into practice.

Patient Ethnicity and Pediatric Visits to the Emergency Department for Fever.

OBJECTIVES: Previous research has identified ethnic differences in parents' beliefs about fever, but whether patient ethnicity is associated with health care use for fever is uncertain. Our objectives were to describe the national rate of pediatric visits to the emergency department (ED) for fever and to determine whether there is variation in this rate by patient ethnicity. METHODS: Using the National Hospital Ambulatory Medical Care Survey between 2012 and 2015, we estimated the proportion of ED visits with a complaint of fever by patients 0 to 18 years old and compared this proportion across patient ethnicity. We performed multivariable logistic regression controlling for sociodemographic characteristics and visit acuity to determine whether patient ethnicity was independently associated with visits for fever. RESULTS: Fever was the reason for 19% [95% confidence interval (CI), 18%-20%] of pediatric visits to the ED, and the proportion of visits for fever was highest among Hispanic patients (25%; 95% CI, 23%-27%) and lowest among non-Hispanic white patients (15%; 95% CI, 14%-17%). In multivariable analysis, the adjusted odds of visits for fever were greater for Hispanic patients (odds ratio, 1.56; 95% CI, 1.38-1.83) and non-Hispanic non-black patients of other races (1.34; 95% CI, 1.02-1.77) compared with non-Hispanic white patients. CONCLUSIONS: There is significant ethnic variation in the use of emergency medical services for fever in the United States, and these disparities are not fully explained by differences in the acuity of illness or differences in socioeconomic status. Interventions to empower parents to manage nonurgent pediatric fever should incorporate ethnocultural differences in parents' understanding of fever.

Seasonality of Common Pediatric Infectious Diseases.

BACKGROUND/OBJECTIVE: Traditional sources cite seasonal patterns for common infectious diseases, often based on microbiologic data, but little is known about cyclical trends in clinically diagnosed infectious conditions in the emergency department (ED). We leveraged the publicly available Nationwide Emergency Department Sample database to measure the seasonality of the most common pediatric infectious diseases diagnosed in US EDs. METHODS: We searched the Nationwide Emergency Department Sample database to identify infectious diagnoses comprising at least 1% of all diagnosis codes ascribed to patients 21 years and younger in US EDs from 2009 to 2013. We used Fourier regression to examine seasonal trends in disease and calculated the peak-to-nadir ratio for each infectious condition. RESULTS: Over 20% of pediatric visits during the study period were for infectious conditions. Upper respiratory infection, otitis media, gastroenteritis, urinary tract infection/pyelonephritis, cellulitis/abscess, and pneumonia showed a seasonal pattern that matched trends found in prior regional or microbiologic-based studies. The strongest seasonal trend as measured by goodness of model fit was found in pneumonia (peak-to-nadir incidence ratio of 2.7), followed by otitis media (2.0), cellulitis/abscess (2.0), gastroenteritis (1.6), upper respiratory infection (3.2), and urinary tract infection/pyelonephritis (1.4). Pharyngitis did not show a strong seasonal trend. CONCLUSIONS: Many of the most common pediatric infectious diseases diagnosed in US EDs exhibited seasonal patterns. Large administrative databases can be used to track seasonal disease patterns, with the advantage that they reflect clinician diagnosis beyond microbiologic confirmation. This methodology could aid in resource planning, infection control, and public health educational initiatives.

Attending-Provider Handoffs and Pediatric Emergency Department Revisits.

OBJECTIVE: The objective of this study was to determine if intradepartment attending-provider transitions of care (handoffs) during a pediatric emergency department (ED) encounter were associated with return ED visits resulting in hospitalization. METHODS: We analyzed ED encounters for patients younger than 21 years discharged from a single pediatric ED from January 2013 to February 2017. We classified an encounter as having a handoff when the initial attending and discharging attending differed. Our primary outcome was a revisit within 72 hours resulting in hospitalization. Our secondary outcomes were any revisit within 72 hours and revisits resulting in hospitalization with potential deficiencies in care. We compared outcome rates for ED encounters with and without provider handoffs, both with and without adjustment for demographic, clinical, and visit characteristics. RESULTS: Of the 177,350 eligible ED encounters, 1961 (1.1%) had a return visit resulting in hospitalization and 6821 (3.9%) had any return visit. In unadjusted analyses, handoffs were associated with an increased likelihood of a return visit resulting in hospitalization (odds ratio [OR], 1.46; 95% confidence interval [CI], 1.26-1.70) or any return visit (OR, 1.20; 95% CI, 1.10-1.31). However, after adjustment, provider handoffs were not associated with return ED visits resulting in hospitalization (OR, 0.96; 95% CI, 0.81-1.13) or any return ED visits (OR, 1.00; 95% CI, 0.90-1.10). CONCLUSIONS: Provider handoffs in a pediatric ED did not increase the risk of return ED visits or return ED visits with deficiencies in care after adjustment for demographic, clinical, and visit factors.

Identifying Patients at Lowest Risk for Streptococcal Pharyngitis: A National Validation Study.

OBJECTIVES: To determine the prevalence of features of viral illness in a national sample of visits involving children tested for group A Streptococcus pharyngitis. Additionally, we sought to derive a decision rule to identify patients with features of viral illness who were at low risk of having group A Streptococcus and for whom laboratory testing might be avoided. STUDY DESIGN: Retrospective validation study using data from electronic health records of patients 3-21 years old evaluated for sore throat in a national network of retail health clinics (n = 67 127). We determined the prevalence of features of viral illness in patients tested for group A Streptococcus and developed a decision tree algorithm to identify patients with features of viral illness at low risk (<15%) of having group A Streptococcus. RESULTS: Overall, 54% of patients had features of viral illness. Among patients with features of viral illness, those without tonsillar exudates who were 11 years or older and either lacked cervical adenopathy or had cervical adenopathy and lacked fever were identified as at low risk for group A Streptococcus according to the decision rule. This group comprised 34% of patients with features of viral illness, or 19% of all patients tested for group A Streptococcus infection. CONCLUSIONS: Our findings provide an objective way to identify patients with features of viral illness who are at low risk of having group A Streptococcus. Improved identification such patients at low risk of group A Streptococcus could improve appropriate testing and antibiotic prescribing for pharyngitis.

Impact of viral symptoms on the performance of the modified centor score to predict pediatric group A streptococcal pharyngitis.

BACKGROUND: Clinicians use the Modified Centor Score (MCS) to estimate the risk of group A streptococcal (GAS) pharyngitis in children with sore throat. The Infectious Diseases Society of America (IDSA) recommends neither testing nor treating patients with specific viral symptoms. The goal of this study is to measure the impact of those symptoms on the yield of GAS testing predicted by the MCS. METHODS: Retrospective cohort study of all patients aged 3-21 years presenting with sore throat and tested for GAS in a pediatric emergency department (ED) in 2016. After identifying all patients tested for GAS, we used natural language processing (NLP) to identify the subgroup complaining of sore throat. We abstracted all MCS variables as well as symptoms suggestive of a viral etiology per the IDSA guideline (conjunctivitis, coryza, cough, diarrhea, hoarseness, ulcerative oral lesions, viral exanthema). We calculated the proportion of patients who tested positive for GAS by MCS with and without viral symptoms. RESULTS: Of the 1574 patients included, 372 patients (24%) tested GAS positive. Patients with at least one viral symptom had a reduced GAS risk compared to those without any of the viral symptoms 91/547 (17% GAS positive) vs. 281/1027 (27%), odds ratio 0.53 (95% CI 0.41-0.69). CONCLUSIONS: The presence of viral symptoms specified by the IDSA alters the predicted yield of testing by traditional MCS. Clinicians may consider adjusting interpretation of a patient's MCS based on the presence of viral symptoms, but viral symptoms may not always fully obviate the need for GAS testing.

Complexity and Severity of Pediatric Patients Treated at United States Emergency Departments.

OBJECTIVE: To compare the complexity and severity of presentation of children in general vs pediatric emergency departments (EDs). STUDY DESIGN: We performed a cross-sectional study of pediatric ED visits using the National Emergency Department Sample from 2008 to 2012. We classified EDs as "pediatric" if >75% of patients were <18 years old; all other EDs were classified as "general." The presence of an International Classification of Diseases, Ninth Revision code for a complex chronic condition was used as an indicator of patient complexity. Patient severity was evaluated with the severity classification system. In addition, rates of critical procedures and hospitalization were assessed. RESULTS: We identified 9.6 million encounters to pediatric EDs and 169 million to general EDs. Younger children account for a greater proportion of visits at pediatric EDs than general EDs; children <1 year of age account for 18% of visits to a pediatric ED compared with 9% of visits to a general ED (P < .01). Encounters at pediatric EDs had greater complexity (5% vs 2%; P < .01). Although severity classification system scores did not significantly differ by ED type, pediatric EDs had greater rates of hospitalization (10% vs 4%). CONCLUSIONS: Pediatric EDs provided care to a greater proportion of medically complex children than general EDs and had greater rates of hospitalization. This information may inform educational efforts in residency or postgraduate training to ensure high-quality care for children with complex health care needs.

Participatory medicine: A home score for streptococcal pharyngitis enabled by real-time biosurveillance: a cohort study.

BACKGROUND: Consensus guidelines recommend against testing or treating adults at low risk for group A streptococcal (GAS) pharyngitis. OBJECTIVE: To help patients decide when to visit a clinician for the evaluation of sore throat. DESIGN: Retrospective cohort study. SETTING: A national chain of retail health clinics. PATIENTS: 71 776 patients aged 15 years or older with pharyngitis who visited a clinic from September 2006 to December 2008. MEASUREMENTS: The authors created a score using information from patient-reported clinical variables plus the incidence of local disease and compared it with the Centor score and other traditional scores that require clinician-elicited signs. RESULTS: If patients aged 15 years or older with sore throat did not visit a clinician when the new score estimated the likelihood of GAS pharyngitis to be less than 10% instead of having clinicians manage their symptoms following guidelines that use the Centor score, 230 000 visits would be avoided in the United States each year and 8500 patients with GAS pharyngitis who would have received antibiotics would not be treated with them. LIMITATION: Real-time information about the local incidence of GAS pharyngitis, which is necessary to calculate the new score, is not currently available. CONCLUSION: A patient-driven approach to pharyngitis diagnosis that uses this new score could save hundreds of thousands of visits annually by identifying patients at home who are unlikely to require testing or treatment. PRIMARY FUNDING SOURCE: Centers for Disease Control and Prevention and the National Library of Medicine, National Institutes of Health.

Impact of a Bronchiolitis Clinical Pathway on Management Decisions by Preferred Language.

Background: Clinical pathways standardize healthcare utilization, but their impact on healthcare equity is poorly understood. This study aims to measure the effect of a bronchiolitis pathway on management decisions by preferred language for care. Methods: We included all emergency department encounters for patients aged 1-12 months with bronchiolitis from 1/1/2010 to 10/31/2020. The prepathway period ended 10/31/2011, and the postpathway period was 1/1/2012-10/31/2020. We performed retrospective interrupted time series analyses to assess the impact of the clinical pathway by English versus non-English preferred language on the following outcomes: chest radiography (CXR), albuterol use, 7-day return visit, 72-hour return to admission, antibiotic use, and corticosteroid use. Analyses were adjusted for presence of a complex chronic condition. Results: There were 1485 encounters in the preperiod (77% English, 14% non-English, 8% missing) and 7840 encounters in the postperiod (79% English, 15% non-English, 6% missing). CXR, antibiotic, and albuterol utilization exhibited sustained decreases over the study period. Pathway impact did not differ by preferred language for any outcome except albuterol utilization. The prepost slope effect of albuterol utilization was 10% greater in the non-English versus the English group (p for the difference by language = 0.022). Conclusions: A clinical pathway was associated with improvements in care regardless of preferred language. More extensive studies involving multiple pathways and care settings are needed to assess the impact of clinical pathways on health equity.

Validation of anaphylaxis in the Food and Drug Administration's Mini-Sentinel.

PURPOSE: We aim to develop and validate the positive predictive value (PPV) of an algorithm to identify anaphylaxis using health plan administrative and claims data. Previously published PPVs for anaphylaxis using International Classification of Diseases, ninth revision, Clinical Modification (ICD-9-CM) codes range from 52% to 57%. METHODS: We conducted a retrospective study using administrative and claims data from eight health plans. Using diagnosis and procedure codes, we developed an algorithm to identify potential cases of anaphylaxis from the Mini-Sentinel Distributed Database between January 2009 and December 2010. A random sample of medical charts (n = 150) was identified for chart abstraction. Two physician adjudicators reviewed each potential case. Using physician adjudicator judgments on whether the case met diagnostic criteria for anaphylaxis, we calculated a PPV for the algorithm. RESULTS: Of the 122 patients for whom complete charts were received, 77 were judged by physician adjudicators to have anaphylaxis. The PPV for the algorithm was 63.1% (95%CI: 53.9-71.7%), using the clinical criteria by Sampson as the gold standard. The PPV was highest for inpatient encounters with ICD-9-CM codes of 995.0 or 999.4. By combining only the top performing ICD-9-CM codes, we identified an algorithm with a PPV of 75.0%, but only 66% of cases of anaphylaxis were identified using this modified algorithm. CONCLUSIONS: The PPV for the ICD-9-CM-based algorithm for anaphylaxis was slightly higher than PPV estimates reported in prior studies, but remained low. We were able to identify an algorithm that optimized the PPV but demonstrated lower sensitivity for anaphylactic events.

Improved diagnostic accuracy of group A streptococcal pharyngitis with use of real-time biosurveillance.

BACKGROUND: Clinical prediction rules do not incorporate real-time incidence data to adjust estimates of disease risk in symptomatic patients. OBJECTIVE: To measure the value of integrating local incidence data into a clinical decision rule for diagnosing group A streptococcal (GAS) pharyngitis in patients aged 15 years or older. DESIGN: Retrospective analysis of clinical and biosurveillance predictors of GAS pharyngitis. SETTING: Large U.S.-based retail health chain. PATIENTS: 82 062 patient visits for pharyngitis. MEASUREMENTS: Accuracy of the Centor score was compared with that of a biosurveillance-responsive score, which was essentially an adjusted Centor score based on real-time GAS pharyngitis information from the 14 days before a patient's visit: the recent local proportion positive (RLPP). RESULTS: Increased RLPP correlated with the likelihood of GAS pharyngitis (r(2) = 0.79; P < 0.001). Local incidence data enhanced diagnostic models. For example, when the RLPP was greater than 0.30, managing patients with Centor scores of 1 as if the scores were 2 would identify 62, 537 previously missed patients annually while misclassifying 18, 446 patients without GAS pharyngitis. Decreasing the score of patients with Centor values of 3 by 1 point for an RLPP less than 0.20 would spare unnecessary antibiotics for 166, 616 patients while missing 18, 812 true-positive cases. LIMITATIONS: Analyses were conducted retrospectively. Real-time regional data on GAS pharyngitis are generally not yet available to clinicians. CONCLUSION: Incorporating live biosurveillance data into clinical guidelines for GAS pharyngitis and other communicable diseases should be considered for reducing missed cases when the contemporaneous incidence is elevated and for sparing unnecessary antibiotics when the contemporaneous incidence is low. Delivering epidemiologic data to the point of care will enable the use of real-time pretest probabilities in medical decision making.

Prediction across healthcare settings: a case study in predicting emergency department disposition.

Several approaches exist today for developing predictive models across multiple clinical sites, yet there is a lack of comparative data on their performance, especially within the context of EHR-based prediction models. We set out to provide a framework for prediction across healthcare settings. As a case study, we examined an ED disposition prediction model across three geographically and demographically diverse sites. We conducted a 1-year retrospective study, including all visits in which the outcome was either discharge-to-home or hospitalization. Four modeling approaches were compared: a ready-made model trained at one site and validated at other sites, a centralized uniform model incorporating data from all sites, multiple site-specific models, and a hybrid approach of a ready-made model re-calibrated using site-specific data. Predictions were performed using XGBoost. The study included 288,962 visits with an overall admission rate of 16.8% (7.9-26.9%). Some risk factors for admission were prominent across all sites (e.g., high-acuity triage emergency severity index score, high prior admissions rate), while others were prominent at only some sites (multiple lab tests ordered at the pediatric sites, early use of ECG at the adult site). The XGBoost model achieved its best performance using the uniform and site-specific approaches (AUC = 0.9-0.93), followed by the calibrated-model approach (AUC = 0.87-0.92), and the ready-made approach (AUC = 0.62-0.85). Our results show that site-specific customization is a key driver of predictive model performance.

Prediction of patient disposition: comparison of computer and human approaches and a proposed synthesis.

OBJECTIVE: To compare the accuracy of computer versus physician predictions of hospitalization and to explore the potential synergies of hybrid physician-computer models. MATERIALS AND METHODS: A single-center prospective observational study in a tertiary pediatric hospital in Boston, Massachusetts, United States. Nine emergency department (ED) attending physicians participated in the study. Physicians predicted the likelihood of admission for patients in the ED whose hospitalization disposition had not yet been decided. In parallel, a random-forest computer model was developed to predict hospitalizations from the ED, based on data available within the first hour of the ED encounter. The model was tested on the same cohort of patients evaluated by the participating physicians. RESULTS: 198 pediatric patients were considered for inclusion. Six patients were excluded due to incomplete or erroneous physician forms. Of the 192 included patients, 54 (28%) were admitted and 138 (72%) were discharged. The positive predictive value for the prediction of admission was 66% for the clinicians, 73% for the computer model, and 86% for a hybrid model combining the two. To predict admission, physicians relied more heavily on the clinical appearance of the patient, while the computer model relied more heavily on technical data-driven features, such as the rate of prior admissions or distance traveled to hospital. DISCUSSION: Computer-generated predictions of patient disposition were more accurate than clinician-generated predictions. A hybrid prediction model improved accuracy over both individual predictions, highlighting the complementary and synergistic effects of both approaches. CONCLUSION: The integration of computer and clinician predictions can yield improved predictive performance.