The effect of computerised decision support alerts tailored to intensive care on the administration of high-risk drug combinations, and their monitoring: a cluster randomised stepped-wedge trial.

BACKGROUND: Drug-drug interactions (DDIs) can harm patients admitted to the intensive care unit (ICU). Yet, clinical decision support systems (CDSSs) aimed at helping physicians prevent DDIs are plagued by low-yield alerts, causing alert fatigue and compromising patient safety. The aim of this multicentre study was to evaluate the effect of tailoring potential DDI alerts to the ICU setting on the frequency of administered high-risk drug combinations. METHODS: We implemented a cluster randomised stepped-wedge trial in nine ICUs in the Netherlands. Five ICUs already used potential DDI alerts. Patients aged 18 years or older admitted to the ICU with at least two drugs administered were included. Our intervention was an adapted CDSS, only providing alerts for potential DDIs considered as high risk. The intervention was delivered at the ICU level and targeted physicians. We hypothesised that showing only relevant alerts would improve CDSS effectiveness and lead to a decreased number of administered high-risk drug combinations. The order in which the intervention was implemented in the ICUs was randomised by an independent researcher. The primary outcome was the number of administered high-risk drug combinations per 1000 drug administrations per patient and was assessed in all included patients. This trial was registered in the Netherlands Trial Register (identifier NL6762) on Nov 26, 2018, and is now closed. FINDINGS: In total, 10 423 patients admitted to the ICU between Sept 1, 2018, and Sept 1, 2019, were assessed and 9887 patients were included. The mean number of administered high-risk drug combinations per 1000 drug administrations per patient was 26·2 (SD 53·4) in the intervention group (n=5534), compared with 35·6 (65·0) in the control group (n=4353). Tailoring potential DDI alerts to the ICU led to a 12% decrease (95% CI 5-18%; p=0·0008) in the number of administered high-risk drug combinations per 1000 drug administrations per patient, after adjusting for clustering and prognostic factors. INTERPRETATION: This cluster randomised stepped-wedge trial showed that tailoring potential DDI alerts to the ICU setting significantly reduced the number of administered high-risk drug combinations. Our list of high-risk drug combinations can be used in other ICUs, and our strategy of tailoring alerts based on clinical relevance could be applied to other clinical settings. FUNDING: ZonMw.

Using the PharmCAT tool for Pharmacogenetic clinical decision support.

Here, we will provide our insights into the usage of PharmCAT as part of a pharmacogenetic clinical decision support pipeline, which addresses the challenges in mapping clinical dosing guidelines to variants to be extracted from genetic datasets. After a general outline of pharmacogenetics, we describe some features of PharmCAT and how we integrated it into a pharmacogenetic clinical decision support system within a clinical information system. We conclude with promising developments regarding future PharmCAT releases.

Implementing a clinical decision support tool to increase early peanut introduction guidance.

BACKGROUND: General pediatric providers are the front line for early peanut introduction discussions, but many providers believe that they are ill-equipped to handle such discussions, as the guidelines have changed quickly. OBJECTIVE: We hypothesized that a clinical decision support (CDS) tool could improve discussions of peanut introduction. METHODS: CDS tools were designed by stakeholders, improved through usability testing, and integrated into the current note templates. On the basis of queries of electronic health records, we did a preperformance versus postperformance evaluation of conversations regarding peanut introduction, barriers to peanut introduction, and percentage of 12-month well-child checkups (WCCs) that resulted in successful introduction of peanut. Providers completed surveys before and after intervention to assess their awareness of early peanut introduction and comfort using the CDS tools. RESULTS: Providers' awareness of early peanut introduction guidelines increased from 17.8% to 66.7% after the CDS tool was implemented; 79.1% of the providers were comfortable using the tool. The CDS tool improved peanut introduction conversations at the 4-month WCC from 2.4% to 81.2%, at the 6-month WCC from 3.0% to 84.2%, and at the 12-month WCC from 2.7% to 82.9%. In all, 56.6% of families had a plan to introduce peanut at the 4-month WCC. Of those who did not have a plan, the most common barrier was the family's unawareness of the benefits of early peanut introduction. At the 12-month WCC, 62.8% of families had introduced peanut without concerns. CONCLUSION: A point-of-care CDS tool encouraged more discussions of early peanut introduction between general pediatric providers and all patients. CDS tools should be considered in quality improvement projects as an implementation method for the most up-to-date guidelines.

Technology in the management of diabetes in hospitalised adults.

Suboptimal glycaemic management in hospitals has been associated with adverse clinical outcomes and increased financial costs to healthcare systems. Despite the availability of guidelines for inpatient glycaemic management, implementation remains challenging because of the increasing workload of clinical staff and rising prevalence of diabetes. The development of novel and innovative technologies that support the clinical workflow and address the unmet need for effective and safe inpatient diabetes care delivery is still needed. There is robust evidence that the use of diabetes technology such as continuous glucose monitoring and closed-loop insulin delivery can improve glycaemic management in outpatient settings; however, relatively little is known of its potential benefits and application in inpatient diabetes management. Emerging data from clinical studies show that diabetes technologies such as integrated clinical decision support systems can potentially mediate safer and more efficient inpatient diabetes care, while continuous glucose sensors and closed-loop systems show early promise in improving inpatient glycaemic management. This review aims to provide an overview of current evidence related to diabetes technology use in non-critical care adult inpatient settings. We highlight existing barriers that may hinder or delay implementation, as well as strategies and opportunities to facilitate the clinical readiness of inpatient diabetes technology in the future.

Computerized Decision Support Systems Informing Community-Acquired Pneumonia Surveillance, France, 2017-2023.

We show the value of real-time data generated by a computerized decision support system in primary care in strengthening pneumonia surveillance. The system showed a 66% (95% CI 64%-67%) increase in community-acquired pneumonia from 2018 to 2023 for the population of France, 1 month before a national alert was issued.

Retrospective evaluation of clinical decision support for within-laboratory optimization of SARS-CoV-2 NAAT workflow.

The unprecedented demand for severe acute respiratory syndrome coronavirus 2 (SARS­CoV­2) testing led to challenges in prioritizing and processing specimens efficiently. We describe and evaluate a novel workflow using provider- and patient-facing ask at order entry (AOE) questions to generate distinctive icons on specimen labels for within-laboratory clinical decision support (CDS) for specimen triaging. A multidisciplinary committee established target turnaround times (TATs) for SARS-CoV-2 nucleic acid amplification test (NAAT) based on common clinical scenarios. A set of AOE questions was used to collect relevant clinical information that prompted icon generation for triaging SARS-CoV-2 NAAT specimens. We assessed the collect-to-verify TATs among relevant clinical scenarios. Our study included a total of 1,385,813 SARS-CoV-2 NAAT conducted from March 2020 to June 2022. Most testing met the TAT targets established by institutional committees, but deviations from target TATs occurred during periods of high demand and supply shortages. Median TATs for emergency department (ED) and inpatient specimens and ambulatory pre-procedure populations were stable over the pandemic. However, healthcare worker and other ambulatory test TATs varied substantially, depending on testing volume and community transmission rates. Median TAT significantly differed throughout the pandemic for ED and inpatient clinical scenarios, and there were significant differences in TAT among label icon-signified ambulatory clinical scenarios. We describe a novel approach to CDS for triaging specimens within the laboratory. The use of CDS tools could help clinical laboratories prioritize and process specimens efficiently, especially during times of high demand. Further studies are needed to evaluate the impact of our CDS tool on overall laboratory efficiency and patient outcomes. IMPORTANCE We describe a novel approach to clinical decision support (CDS) for triaging specimens within the clinical laboratory for severe acute respiratory syndrome coronavirus 2 (SARS­CoV­2) nucleic acid amplification tests (NAAT). The use of our CDS tool could help clinical laboratories prioritize and process specimens efficiently, especially during times of high demand. There were significant differences in the turnaround time for specimens differentiated by icons on specimen labels. Further studies are needed to evaluate the impact of our CDS tool on overall laboratory efficiency and patient outcomes.

Rationale, Design and Baseline Characteristics of a Randomized Controlled Trial of a Cardiovascular Quality Improvement Strategy in India: The C-QIP Trial.

BACKGROUND: Quality of chronic care for cardiovascular disease (CVD) remains suboptimal worldwide. The Collaborative Quality ImProvement (C-QIP) trial aims to develop and test the feasibility and clinical effect of a multicomponent strategy among patients with prevalent CVD in India. METHODS: The C-QIP is a clinic-based, open randomized trial of a multicomponent intervention vs usual care that was locally developed and adapted for use in Indian settings through rigorous formative research guided by Consolidated Framework for Implementation Research (CFIR). The C-QIP intervention consisted of 5 components: 1) electronic health records and decision support system for clinicians, 2) trained nonphysician health workers (NPHW), 3) text-message based lifestyle reminders, 4) patient education materials, 5) quarterly audit and feedback reports. Patients with CVD (ischemic heart disease, ischemic stroke, or heart failure) attending outpatient CVD clinics were recruited from September 2022 to September 2023 and were randomized to the intervention or usual care arm for at least 12 months follow-up. The co-primary outcomes are implementation feasibility, fidelity (ie, dose delivered and dose received), acceptability, adoption and appropriateness, measured at multiple levels: patient, provider and clinic site-level, The secondary outcomes include prescription of guideline directed medical therapy (GDMT) (provider-level), and adherence to prescribed therapy, change in mean blood pressure (BP) and LDL-cholesterol between the intervention and control groups (patient-level). In addition, a trial-based process and economic evaluations will be performed using standard guidelines. RESULTS: We recruited 410 socio-demographically diverse patients with CVD from 4 hospitals in India. Mean (SD) age was 57.5 (11.7) years, and 73.0% were males. Self-reported history of hypertension (48.5%) and diabetes (41.5%) was common. At baseline, mean (SD) BP was 127.9 (18.2) /76.2 (11.6) mm Hg, mean (SD) LDLc: 80.3 (37.3) mg/dl and mean (SD) HbA1c: 6.8% (1.6%). At baseline, the GDMT varied from 62.4% for patients with ischemic heart disease, 48.6% for ischemic stroke and 36.1% for heart failure. CONCLUSION: This study will establish the feasibility of delivering contextually relevant, and evidence-based C-QIP strategy and assess whether it is acceptable to the target populations. The study results will inform a larger scale confirmatory trial of a comprehensive CVD care model in low-resource settings. TRIAL REGISTRATION: Clinical Trials Registry India: CTRI/2022/04/041847; Clinicaltrials.gov number: NCT05196659.

Design and rationale of the Comprehensive intelligent Hypertension managEment SyStem (CHESS) evaluation study: A cluster randomized controlled trial for hypertension management in primary care.

BACKGROUND: Hypertension management in China is suboptimal with high prevalence and low control rate due to various barriers, including lack of self-management awareness of patients and inadequate capacity of physicians. Digital therapeutic interventions including mobile health and computational device algorithms such as clinical decision support systems (CDSS) are scalable with the potential to improve blood pressure (BP) management and strengthen the healthcare system in resource-constrained areas, yet their effectiveness remains to be tested. The aim of this report is to describe the protocol of the Comprehensive intelligent Hypertension managEment SyStem (CHESS) evaluation study assessing the effect of a multifaceted hypertension management system for supporting patients and physicians on BP lowering in primary care settings. MATERIALS AND METHODS: The CHESS evaluation study is a parallel-group, cluster-randomized controlled trial conducted in primary care settings in China. Forty-one primary care sites from 3 counties of China are randomly assigned to either the usual care or the intervention group with the implementation of the CHESS system, more than 1,600 patients aged 35 to 80 years with uncontrolled hypertension and access to a smartphone by themselves or relatives are recruited into the study and followed up for 12 months. In the intervention group, participants receive patient-tailored reminders and alerts via messages or intelligent voice calls triggered by uploaded home blood pressure monitoring data and participants' characteristics, while physicians receive guideline-based prescription instructions according to updated individual data from each visit, and administrators receive auto-renewed feedback of hypertension management performance from the data analysis platform. The multiple components of the CHESS system can work synergistically and have undergone rigorous development and pilot evaluation using a theory-informed approach. The primary outcome is the mean change in 24-hour ambulatory systolic BP from baseline to 12 months. DISCUSSION: The CHESS trial will provide evidence and novel insight into the effectiveness and feasibility of an implementation strategy using a comprehensive digital BP management system for reducing hypertension burden in primary care settings. TRIAL REGISTRATION: https://www. CLINICALTRIALS: gov, NCT05605418.

Computerized clinical decision support to improve stroke prevention therapy in primary care management of atrial fibrillation: a cluster randomized trial.

BACKGROUND: Despite guidelines supporting antithrombotic therapy use in atrial fibrillation (AF), under-prescribing persists. We assessed whether computerized clinical decision support (CDS) would enable guideline-based antithrombotic therapy for AF patients in primary care. METHODS: This cluster randomized trial of CDS versus usual care (UC) recruited participants from primary care practices across Nova Scotia, following them for 12 months. The CDS tool calculated bleeding and stroke risk scores and provided recommendations for using oral anticoagulants (OAC) per Canadian guidelines. RESULTS: From June 14, 2014 to December 15, 2016, 203 primary care providers (99 UC, 104 CDS) with access to high-speed Internet were recruited, enrolling 1,145 eligible patients (543 UC, 590 CDS) assigned to the same treatment arm as their provider. Patient mean age was 72.3 years; most were male (350, 64.5% UC, 351, 59.5% CDS) and from a rural area (298, 54.9% UC, 315, 53.4% CDS). At baseline, a higher than anticipated proportion of patients were receiving guideline-based OAC therapy (373, 68.7% UC, 442, 74.9% CDS; relative risk [RR] 0.97 (95% confidence interval [CI], 0.87-1.07; P = .511)). At 12 months, prescription data were available for 538 usual care and 570 CDS patients, and significantly more CDS patients were managed according to guidelines (415, 77.1% UC, 479, 84.0% CDS; RR 1.08 (95% CI, 1.01-1.15; P = .024)). CONCLUSION: Notwithstanding high baseline rates, primary care provider access to the CDS over 12 months further optimized the prescribing of OAC therapy per national guidelines to AF patients potentially eligible to receive it. This suggests that CDS can be effective in improving clinical process of care. TRIAL REGISTRATION: Clinical Trials NCT01927367. https://clinicaltrials.gov/ct2/show/NCT01927367?term=NCT01927367&draw=2&rank=1.

ICU Utilization After Implementation of Minor Severe Pneumonia Criteria in Real-Time Electronic Clinical Decision Support.

OBJECTIVES: To determine if the implementation of automated clinical decision support (CDS) with embedded minor severe community-acquired pneumonia (sCAP) criteria was associated with improved ICU utilization among emergency department (ED) patients with pneumonia who did not require vasopressors or positive pressure ventilation at admission. DESIGN: Planned secondary analysis of a stepped-wedge, cluster-controlled CDS implementation trial. SETTING: Sixteen hospitals in six geographic clusters from Intermountain Health; a large, integrated, nonprofit health system in Utah and Idaho. PATIENTS: Adults admitted to the hospital from the ED with pneumonia identified by: 1) discharge International Classification of Diseases , 10th Revision codes for pneumonia or sepsis/respiratory failure and 2) ED chest imaging consistent with pneumonia, who did not require vasopressors or positive pressure ventilation at admission. INTERVENTIONS: After implementation, patients were exposed to automated, open-loop, comprehensive CDS that aided disposition decision (ward vs. ICU), based on objective severity scores (sCAP). MEASUREMENTS AND MAIN RESULTS: The analysis included 2747 patients, 1814 before and 933 after implementation. The median age was 71, median Elixhauser index was 17, 48% were female, and 95% were Caucasian. A mixed-effects regression model with cluster as the random effect estimated that implementation of CDS utilizing sCAP increased 30-day ICU-free days by 1.04 days (95% CI, 0.48-1.59; p < 0.001). Among secondary outcomes, the odds of being admitted to the ward, transferring to the ICU within 72 hours, and receiving a critical therapy decreased by 57% (odds ratio [OR], 0.43; 95% CI, 0.26-0.68; p < 0.001) post-implementation; mortality within 72 hours of admission was unchanged (OR, 1.08; 95% CI, 0.56-2.01; p = 0.82) while 30-day all-cause mortality was lower post-implementation (OR, 0.71; 95% CI, 0.52-0.96; p = 0.03). CONCLUSIONS: Implementation of electronic CDS using minor sCAP criteria to guide disposition of patients with pneumonia from the ED was associated with safe reduction in ICU utilization.

Correlation between antifungal clinical practices and a new clinical decision support system ANTIFON-CLIC® for the treatment of invasive candidiasis: a retrospective multicentre study.

BACKGROUND: Invasive candidiasis is still recognized as a major cause of morbidity and mortality. To support clinicians in the optimal use of antifungals for the treatment of invasive candidiasis, a computerized decision support system (CDSS) was developed based on institutional guidelines. OBJECTIVES: To evaluate the correlation of this newly developed CDSS with clinical practices, we set-up a retrospective multicentre cohort study with the aim of providing the concordance rate between the CDSS recommendation and the medical prescription (NCT05656157). PATIENTS AND METHODS: Adult patients who received caspofungin or fluconazole for the treatment of an invasive candidiasis were included. The analysis of factors associated with concordance was performed using mixed logistic regression models with department as a random effect. RESULTS: From March to November 2022, 190 patients were included from three centres and eight departments: 70 patients from centre A, 84 from centre B and 36 from centre C. Overall, 100 patients received caspofungin and 90 received fluconazole, mostly (59%; 112/190) for empirical/pre-emptive treatment. The overall percentage of concordance between the CDSS and medical prescriptions was 91% (173/190) (confidence interval 95%: 82%-96%). No significant difference in concordance was observed considering the centres (P > 0.99), the department of inclusion (P = 0.968), the antifungal treatment (P = 0.656) or the indication of treatment (P = 0.997). In most cases of discordance (n = 13/17, 76%), the CDSS recommended fluconazole whereas caspofungin was prescribed. The clinical usability evaluated by five clinicians was satisfactory. CONCLUSIONS: Our results demonstrated the high correlation between current antifungal clinical practice and this user-friendly and institutional guidelines-based CDSS.

Documentation of results and medication prescribing after combinatorial psychiatric pharmacogenetic testing: A case for discrete results.

PURPOSE: Combinatorial pharmacogenetic (PGx) panels intended to aid psychiatric prescribing are available to clinicians. Here, we evaluated the documentation of PGx panel results and subsequent prescribing patterns within a tertiary health care system. METHODS: We performed a query of psychiatry service note text in our electronic health record using 71 predefined PGx terms. Patients who underwent combinatorial PGx testing were identified, and documentation of test results was analyzed. Prescription data following testing were examined for the frequency of prescriptions influenced by genes on the panel along with the medical specialties involved. RESULTS: A total of 341 patients received combinatorial PGx testing, and documentation of results was found to be absent or incomplete for 198 patients (58%). The predominant method of documentation was through portable document formats uploaded to the electronic health record's "Media" section. Among patients with at least 1 year of follow-up, a large majority (194/228, 85%) received orders for medications affected by the tested genes, including 132 of 228 (58%) patients receiving at least 1 non-psychiatric medication influenced by the test results. CONCLUSION: Results from combinatorial PGx testing were poorly documented. Medications affected by these results were often prescribed after testing, highlighting the need for discrete results and clinical decision support.

Comparison of Clinical Decision Support Tools to Improve Pediatric Lipid Screening.

OBJECTIVE: To test whether different clinical decision support tools increase clinician orders and patient completions relative to standard practice and each other. STUDY DESIGN: A pragmatic, patient-randomized clinical trial in the electronic health record was conducted between October 2019 and April 2020 at Geisinger Health System in Pennsylvania, with 4 arms: care gap-a passive listing recommending screening; alert-a panel promoting and enabling lipid screen orders; both; and a standard practice-no guideline-based notification-control arm. Data were analyzed for 13 346 9- to 11-year-old patients seen within Geisinger primary care, cardiology, urgent care, or nutrition clinics, or who had an endocrinology visit. Principal outcomes were lipid screening orders by clinicians and completions by patients within 1 week of orders. RESULTS: Active (care gap and/or alert) vs control arm patients were significantly more likely (P < .05) to have lipid screening tests ordered and completed, with ORs ranging from 1.67 (95% CI 1.28-2.19) to 5.73 (95% CI 4.46-7.36) for orders and 1.54 (95% CI 1.04-2.27) to 2.90 (95% CI 2.02-4.15) for completions. Alerts, with or without care gaps listed, outperformed care gaps alone on orders, with odds ratios ranging from 2.92 (95% CI 2.32-3.66) to 3.43 (95% CI 2.73-4.29). CONCLUSIONS: Electronic alerts can increase lipid screening orders and completions, suggesting clinical decision support can improve guideline-concordant screening. The study also highlights electronic record-based patient randomization as a way to determine relative effectiveness of support tools. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT04118348.

Clinical Decision Support for Improved Neonatal Care: The Development of a Machine Learning Model for the Prediction of Late-onset Sepsis and Necrotizing Enterocolitis.

OBJECTIVE: To develop an artificial intelligence-based software system for predicting late-onset sepsis (LOS) and necrotizing enterocolitis (NEC) in infants admitted to the neonatal intensive care unit (NICU). STUDY DESIGN: Single-center, retrospective cohort study, conducted in the NICU of the Antwerp University Hospital. Continuous monitoring data of 865 preterm infants born at <32 weeks gestational age, admitted to the NICU in the first week of life, were used to train an XGBoost machine learning (ML) algorithm for LOS and NEC prediction in a cross-validated setup. Afterward, the model's performance was assessed on an independent test set of 148 patients (internal validation). RESULTS: The ML model delivered hourly risk predictions with an overall sensitivity of 69% (142/206) for all LOS/NEC episodes and 81% (67/83) for severe LOS/NEC episodes. The model showed a median time gain of ≤10 hours (IQR, 3.1-21.0 hours), compared with historical clinical diagnosis. On the complete retrospective dataset, the ML model made 721 069 predictions, of which 9805 (1.3%) depicted a LOS/NEC probability of ≥0.15, resulting in a total alarm rate of <1 patient alarm-day per week. The model reached a similar performance on the internal validation set. CONCLUSIONS: Artificial intelligence technology can assist clinicians in the early detection of LOS and NEC in the NICU, which potentially can result in clinical and socioeconomic benefits. Additional studies are required to quantify further the effect of combining artificial and human intelligence on patient outcomes in the NICU.

Integrating large language models in care, research, and education in multiple sclerosis management.

Use of techniques derived from generative artificial intelligence (AI), specifically large language models (LLMs), offer a transformative potential on the management of multiple sclerosis (MS). Recent LLMs have exhibited remarkable skills in producing and understanding human-like texts. The integration of AI in imaging applications and the deployment of foundation models for the classification and prognosis of disease course, including disability progression and even therapy response, have received considerable attention. However, the use of LLMs within the context of MS remains relatively underexplored. LLMs have the potential to support several activities related to MS management. Clinical decision support systems could help selecting proper disease-modifying therapies; AI-based tools could leverage unstructured real-world data for research or virtual tutors may provide adaptive education materials for neurologists and people with MS in the foreseeable future. In this focused review, we explore practical applications of LLMs across the continuum of MS management as an initial scope for future analyses, reflecting on regulatory hurdles and the indispensable role of human supervision.

Evaluation of an artificial intelligence-based decision support for the detection of cutaneous melanoma in primary care: a prospective real-life clinical trial.

BACKGROUND: Use of artificial intelligence (AI), or machine learning, to assess dermoscopic images of skin lesions to detect melanoma has, in several retrospective studies, shown high levels of diagnostic accuracy on par with - or even outperforming - experienced dermatologists. However, the enthusiasm around these algorithms has not yet been matched by prospective clinical trials performed in authentic clinical settings. In several European countries, including Sweden, the initial clinical assessment of suspected skin cancer is principally conducted in the primary healthcare setting by primary care physicians, with or without access to teledermoscopic support from dermatology clinics. OBJECTIVES: To determine the diagnostic performance of an AI-based clinical decision support tool for cutaneous melanoma detection, operated by a smartphone application (app), when used prospectively by primary care physicians to assess skin lesions of concern due to some degree of melanoma suspicion. METHODS: This prospective multicentre clinical trial was conducted at 36 primary care centres in Sweden. Physicians used the smartphone app on skin lesions of concern by photographing them dermoscopically, which resulted in a dichotomous decision support text regarding evidence for melanoma. Regardless of the app outcome, all lesions underwent standard diagnostic procedures (surgical excision or referral to a dermatologist). After investigations were complete, lesion diagnoses were collected from the patients' medical records and compared with the app's outcome and other lesion data. RESULTS: In total, 253 lesions of concern in 228 patients were included, of which 21 proved to be melanomas, with 11 thin invasive melanomas and 10 melanomas in situ. The app's accuracy in identifying melanomas was reflected in an area under the receiver operating characteristic (AUROC) curve of 0.960 [95% confidence interval (CI) 0.928-0.980], corresponding to a maximum sensitivity and specificity of 95.2% and 84.5%, respectively. For invasive melanomas alone, the AUROC was 0.988 (95% CI 0.965-0.997), corresponding to a maximum sensitivity and specificity of 100% and 92.6%, respectively. CONCLUSIONS: The clinical decision support tool evaluated in this investigation showed high diagnostic accuracy when used prospectively in primary care patients, which could add significant clinical value for primary care physicians assessing skin lesions for melanoma.

Influence of artificial intelligence on the diagnostic performance of endoscopists in the assessment of Barrett's esophagus: a tandem randomized and video trial.

BACKGROUND: This study evaluated the effect of an artificial intelligence (AI)-based clinical decision support system on the performance and diagnostic confidence of endoscopists in their assessment of Barrett's esophagus (BE). METHODS: 96 standardized endoscopy videos were assessed by 22 endoscopists with varying degrees of BE experience from 12 centers. Assessment was randomized into two video sets: group A (review first without AI and second with AI) and group B (review first with AI and second without AI). Endoscopists were required to evaluate each video for the presence of Barrett's esophagus-related neoplasia (BERN) and then decide on a spot for a targeted biopsy. After the second assessment, they were allowed to change their clinical decision and confidence level. RESULTS: AI had a stand-alone sensitivity, specificity, and accuracy of 92.2%, 68.9%, and 81.3%, respectively. Without AI, BE experts had an overall sensitivity, specificity, and accuracy of 83.3%, 58.1%, and 71.5%, respectively. With AI, BE nonexperts showed a significant improvement in sensitivity and specificity when videos were assessed a second time with AI (sensitivity 69.8% [95%CI 65.2%-74.2%] to 78.0% [95%CI 74.0%-82.0%]; specificity 67.3% [95%CI 62.5%-72.2%] to 72.7% [95%CI 68.2%-77.3%]). In addition, the diagnostic confidence of BE nonexperts improved significantly with AI. CONCLUSION: BE nonexperts benefitted significantly from additional AI. BE experts and nonexperts remained significantly below the stand-alone performance of AI, suggesting that there may be other factors influencing endoscopists' decisions to follow or discard AI advice.

Multiobjective tree-based reinforcement learning for estimating tolerant dynamic treatment regimes.

A dynamic treatment regime (DTR) is a sequence of treatment decision rules that dictate individualized treatments based on evolving treatment and covariate history. It provides a vehicle for optimizing a clinical decision support system and fits well into the broader paradigm of personalized medicine. However, many real-world problems involve multiple competing priorities, and decision rules differ when trade-offs are present. Correspondingly, there may be more than one feasible decision that leads to empirically sufficient optimization. In this paper, we propose a concept of "tolerant regime," which provides a set of individualized feasible decision rules under a prespecified tolerance rate. A multiobjective tree-based reinforcement learning (MOT-RL) method is developed to directly estimate the tolerant DTR (tDTR) that optimizes multiple objectives in a multistage multitreatment setting. At each stage, MOT-RL constructs an unsupervised decision tree by modeling the counterfactual mean outcome of each objective via semiparametric regression and maximizing a purity measure constructed by the scalarized augmented inverse probability weighted estimators (SAIPWE). The algorithm is implemented in a backward inductive manner through multiple decision stages, and it estimates the optimal DTR and tDTR depending on the decision-maker's preferences. Multiobjective tree-based reinforcement learning is robust, efficient, easy-to-interpret, and flexible to different settings. We apply MOT-RL to evaluate 2-stage chemotherapy regimes that reduce disease burden and prolong survival for advanced prostate cancer patients using a dataset collected at MD Anderson Cancer Center.

Reader bias in breast cancer screening related to cancer prevalence and artificial intelligence decision support-a reader study.

OBJECTIVES: The aim of our study was to examine how breast radiologists would be affected by high cancer prevalence and the use of artificial intelligence (AI) for decision support. MATERIALS AND METHOD: This reader study was based on selection of screening mammograms, including the original radiologist assessment, acquired in 2010 to 2013 at the Karolinska University Hospital, with a ratio of 1:1 cancer versus healthy based on a 2-year follow-up. A commercial AI system generated an exam-level positive or negative read, and image markers. Double-reading and consensus discussions were first performed without AI and later with AI, with a 6-week wash-out period in between. The chi-squared test was used to test for differences in contingency tables. RESULTS: Mammograms of 758 women were included, half with cancer and half healthy. 52% were 40-55 years; 48% were 56-75 years. In the original non-enriched screening setting, the sensitivity was 61% (232/379) at specificity 98% (323/379). In the reader study, the sensitivity without and with AI was 81% (307/379) and 75% (284/379) respectively (p < 0.001). The specificity without and with AI was 67% (255/379) and 86% (326/379) respectively (p < 0.001). The tendency to change assessment from positive to negative based on erroneous AI information differed between readers and was affected by type and number of image signs of malignancy. CONCLUSION: Breast radiologists reading a list with high cancer prevalence performed at considerably higher sensitivity and lower specificity than the original screen-readers. Adding AI information, calibrated to a screening setting, decreased sensitivity and increased specificity. CLINICAL RELEVANCE STATEMENT: Radiologist screening mammography assessments will be biased towards higher sensitivity and lower specificity by high-risk triaging and nudged towards the sensitivity and specificity setting of AI reads. After AI implementation in clinical practice, there is reason to carefully follow screening metrics to ensure the impact is desired. KEY POINTS: • Breast radiologists' sensitivity and specificity will be affected by changes brought by artificial intelligence. • Reading in a high cancer prevalence setting markedly increased sensitivity and decreased specificity. • Reviewing the binary reads by AI, negative or positive, biased screening radiologists towards the sensitivity and specificity of the AI system.