Shitij Arora
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Shitij Arora has a total of 193 co-authors. Here are the top co-authors by number of shared publications:
- Manasi Agrawal
The Henry D. Janowitz Division of Gastroenterology, Icahn School of Medicine at Mount Sinai, New York, USA
5 shared publications Abhinav Yadav
4 shared publications- Pravin C. Singhal
Feinstein Institute for Medical Research, North Shore-Long Island Jewish Health System, Albert Einstein College of Medicine, Bronx, NY, United States
4 shared publications Yatin Sethi
4 shared publications- Yoram A. Puius
Montefiore Medical Center, Albert Einstein College of Medicine, Bronx, New York, United States
4 shared publications Anju Yadav
Department of Medicine, North Shore and Long Island Jewish Health System and Feinstein Institute for Medical Research, Manhasset, New York, USA
3 shared publications
Latest Publications
- Open Access
- 7 August 2026
- Published byĀ JMIR Publications Inc.Ā inĀ JMIR Preprints
Policy-Anchored Evaluation of Large Language Model Disposition Recommendations for ED Chest Pain: Single-Institution Feasibility Study (RAPID-ED Phase 0) (Preprint)
BACKGROUND Large language models (LLMs) are increasingly deployed as clinical decision support systems (CDSS), but the accuracy of their emergency department (ED) disposition recommendations is unclear. We propose measuring not just accuracy but also direction of error and adherence to institutional protocols. OBJECTIVE To assess the accuracy of policy-anchored LLM disposition recommendations for chest pain presentations to the ED, and to characterize concordance, direction of discordance, and institutional disposition-vocabulary adherence across 5 artificial intelligence (AI) configurations. METHODS In a single-institution feasibility study, 31 ED chest pain cases (21 real, 10 synthetic borderline-risk vignettes) were scored on a 5-tier ordinal disposition scale (discharge, observation, floor, telemetry, intensive care/cath lab). An emergency physician (EP) adjudicating against the institutional chest pain policy served as the reference standard (Reviewer 1) and was compared to a second EPās review to establish ceiling inter-rater reliability. Five AI configurations were asked to provide level of care and campus disposition: an ungrounded general-purpose model (ChatGPT-Edu, GPT-5.5); 3 retrieval-augmented generation (RAG) -like configurations grounded in the 2021 national chest pain guideline and institutional disposition policy, differing in knowledge-delivery prompting; and OpenEvidence, a literature-grounded RAG LLM. The primary metric was linear-weighted Gwet AC2, with 95% CIs by bootstrap. Discordant cases were classified as over- or under-triage; agreement with actual disposition was also measured. RESULTS Physician adjudication agreed on 30/31 cases; inter-rater reliability was AC2 0.98, 95% CI 0.95-1.00. Among LLM configurations, per-case in-chat delivery achieved the highest agreement (AC2 0.53, 95% CI 0.39-0.68) and single-batch delivery the lowest (0.45), with OpenEvidence at 0.49; the ungrounded baseline was lowest overall (0.31, 95% CI 0.10-0.52). Reviewer 1 agreed with actual disposition (n=21) at AC2 0.81 (0.64-0.96), exceeding every configuration. GPT base over-triaged (22 over vs 3 under; telemetry or intensive care in 77% of cases vs 26% for the reference), whereas RAG-batch and OpenEvidence tended to under-triage (6 over vs 14 under). CONCLUSIONS No LLM configuration approached the level of physician-physician agreement. Institutional grounding improved concordance but reversed the direction of residual error, from over-triage to under-triage. This error was driven predominantly by an observation pathway not actually present at the institution. Adherence to the institutional disposition vocabulary is a distinct evaluation axis from accuracy; deployment should constrain model output to the local disposition menu.
- 7 August 2026
- Published byĀ JMIR Publications Inc.Ā inĀ JMIR Preprints
Open AccessPolicy-Anchored Evaluation of Large Language Model Disposition Recommendations for ED Chest Pain: Single-Institution Feasibility Study (RAPID-ED Phase 0) (Preprint)
BACKGROUND Large language models (LLMs) are increasingly deployed as clinical decision support systems (CDSS), but the accuracy of their emergency department (ED) disposition recommendations is unclear. We propose measuring not just accuracy but also direction of error and adherence to institutional protocols. OBJECTIVE To assess the accuracy of policy-anchored LLM disposition recommendations for chest pain presentations to the ED, and to characterize concordance, direction of discordance, and institutional disposition-vocabulary adherence across 5 artificial intelligence (AI) configurations. METHODS In a single-institution feasibility study, 31 ED chest pain cases (21 real, 10 synthetic borderline-risk vignettes) were scored on a 5-tier ordinal disposition scale (discharge, observation, floor, telemetry, intensive care/cath lab). An emergency physician (EP) adjudicating against the institutional chest pain policy served as the reference standard (Reviewer 1) and was compared to a second EPās review to establish ceiling inter-rater reliability. Five AI configurations were asked to provide level of care and campus disposition: an ungrounded general-purpose model (ChatGPT-Edu, GPT-5.5); 3 retrieval-augmented generation (RAG) -like configurations grounded in the 2021 national chest pain guideline and institutional disposition policy, differing in knowledge-delivery prompting; and OpenEvidence, a literature-grounded RAG LLM. The primary metric was linear-weighted Gwet AC2, with 95% CIs by bootstrap. Discordant cases were classified as over- or under-triage; agreement with actual disposition was also measured. RESULTS Physician adjudication agreed on 30/31 cases; inter-rater reliability was AC2 0.98, 95% CI 0.95-1.00. Among LLM configurations, per-case in-chat delivery achieved the highest agreement (AC2 0.53, 95% CI 0.39-0.68) and single-batch delivery the lowest (0.45), with OpenEvidence at 0.49; the ungrounded baseline was lowest overall (0.31, 95% CI 0.10-0.52). Reviewer 1 agreed with actual disposition (n=21) at AC2 0.81 (0.64-0.96), exceeding every configuration. GPT base over-triaged (22 over vs 3 under; telemetry or intensive care in 77% of cases vs 26% for the reference), whereas RAG-batch and OpenEvidence tended to under-triage (6 over vs 14 under). CONCLUSIONS No LLM configuration approached the level of physician-physician agreement. Institutional grounding improved concordance but reversed the direction of residual error, from over-triage to under-triage. This error was driven predominantly by an observation pathway not actually present at the institution. Adherence to the institutional disposition vocabulary is a distinct evaluation axis from accuracy; deployment should constrain model output to the local disposition menu.
- Open Access
Impact of a Smart Phone-Based Educational Video on Bowel Preparation and Colonoscopy Adherence: A Propensity Score Matched Analysis
Background and Aims: Inadequate bowel preparation and high cancellation rates hinder the effectiveness of colonoscopy screening programs. We aimed to evaluate the impact of a novel, personalized video messaging platform on bowel preparation quality and colonoscopy adherence. Methods: We conducted a before-after intervention study at a large urban academic medical center, comparing outcomes between a historical control group (April-September 2022) and an intervention group receiving personalized video messages (July-November 2023). Primary outcomes were inadequate bowel preparation and colonoscopy cancellation/rescheduling rates. Propensity score matching and multivariable logistic regression were used to adjust for potential confounders. Results: Among 1,647 patients who underwent colonoscopy, inadequate bowel preparation occurred in 12.0% of the intervention group versus 20.0% of the historical control group in the propensity score matched analysis (p=0.041). In the larger cohort of 2,802 scheduled patients, the propensity score matched analysis showed a trend toward a 31% reduction in odds of cancellation or rescheduling associated with the educational video, although this did not reach statistical significance (OR 0.69, 95% CI 0.47-1.01, p=0.056). For the logistical regression model, the video was associated with a 42% reduction in odds of cancellation or rescheduling (adjusted OR 0.58, 95% CI 0.44-0.77, p<0.001). Conclusions: Implementation of a personalized video messaging platform was associated with significant improvements in bowel preparation quality and substantial reductions in colonoscopy cancellations/rescheduling. This easily scalable, low-cost intervention has the potential to enhance the efficiency and effectiveness of colonoscopy screening programs.
Open AccessImpact of a Smart Phone-Based Educational Video on Bowel Preparation and Colonoscopy Adherence: A Propensity Score Matched Analysis
Background and Aims: Inadequate bowel preparation and high cancellation rates hinder the effectiveness of colonoscopy screening programs. We aimed to evaluate the impact of a novel, personalized video messaging platform on bowel preparation quality and colonoscopy adherence. Methods: We conducted a before-after intervention study at a large urban academic medical center, comparing outcomes between a historical control group (April-September 2022) and an intervention group receiving personalized video messages (July-November 2023). Primary outcomes were inadequate bowel preparation and colonoscopy cancellation/rescheduling rates. Propensity score matching and multivariable logistic regression were used to adjust for potential confounders. Results: Among 1,647 patients who underwent colonoscopy, inadequate bowel preparation occurred in 12.0% of the intervention group versus 20.0% of the historical control group in the propensity score matched analysis (p=0.041). In the larger cohort of 2,802 scheduled patients, the propensity score matched analysis showed a trend toward a 31% reduction in odds of cancellation or rescheduling associated with the educational video, although this did not reach statistical significance (OR 0.69, 95% CI 0.47-1.01, p=0.056). For the logistical regression model, the video was associated with a 42% reduction in odds of cancellation or rescheduling (adjusted OR 0.58, 95% CI 0.44-0.77, p<0.001). Conclusions: Implementation of a personalized video messaging platform was associated with significant improvements in bowel preparation quality and substantial reductions in colonoscopy cancellations/rescheduling. This easily scalable, low-cost intervention has the potential to enhance the efficiency and effectiveness of colonoscopy screening programs.
- 19 April 2024
- Published byĀ WileyĀ inĀ Journal of Hospital Medicine
- Vol.Ā 19 Ā (12) , 1197-1200
Artificial intelligence in medicine: A primer and recommendation
- 19 April 2024
- Published byĀ WileyĀ inĀ Journal of Hospital Medicine
- Vol.Ā 19 Ā (12) , 1197-1200
Artificial intelligence in medicine: A primer and recommendation
- 10 April 2024
- Published byĀ WileyĀ inĀ Journal of Hospital Medicine
- Vol.Ā 19 Ā (12) , 1170-1173
Things We Do for No Reasonā¢: Routine respiratory pathogen panels for emergency department and hospitalized patients
- 10 April 2024
- Published byĀ WileyĀ inĀ Journal of Hospital Medicine
- Vol.Ā 19 Ā (12) , 1170-1173
Things We Do for No Reasonā¢: Routine respiratory pathogen panels for emergency department and hospitalized patients
- Open Access
- 8 March 2023
- Published byĀ JMIR Publications Inc.Ā inĀ JMIR Public Health and Surveillance
- Vol.Ā 9 Ā (1) , e45121
Technology, Training, and Task Shifting at the Worldās Largest Mass Gathering in 2025: An Opportunity for Antibiotic Stewardship in India
The role of antibiotic overuse in intensifying selection pressures and contributing to the emergence of antimicrobial resistance is well established. The Kumbh Mela, a religious festival that occurs in 4 Indian cities of spiritual significance, is the worldās largest mass gathering, attracting over 80 million pilgrims in 2013. Digital syndromic surveillance from the 2013 and 2015 Melas demonstrated a consistent pattern of antibiotic overuse, with an antibiotic prescribing rate of up to 31% for all patient encounters. As preparations for the 2025 Kumbh Mela begin, task shifting, point-of-care diagnostic and digital tools, robust clinician training, and community awareness can promote the restrained and evidence-based use of antibiotics, minimizing the potential for the emergence of antimicrobial resistance at the worldās largest mass gathering.
- 8 March 2023
- Published byĀ JMIR Publications Inc.Ā inĀ JMIR Public Health and Surveillance
- Vol.Ā 9 Ā (1) , e45121
Open AccessTechnology, Training, and Task Shifting at the Worldās Largest Mass Gathering in 2025: An Opportunity for Antibiotic Stewardship in India
The role of antibiotic overuse in intensifying selection pressures and contributing to the emergence of antimicrobial resistance is well established. The Kumbh Mela, a religious festival that occurs in 4 Indian cities of spiritual significance, is the worldās largest mass gathering, attracting over 80 million pilgrims in 2013. Digital syndromic surveillance from the 2013 and 2015 Melas demonstrated a consistent pattern of antibiotic overuse, with an antibiotic prescribing rate of up to 31% for all patient encounters. As preparations for the 2025 Kumbh Mela begin, task shifting, point-of-care diagnostic and digital tools, robust clinician training, and community awareness can promote the restrained and evidence-based use of antibiotics, minimizing the potential for the emergence of antimicrobial resistance at the worldās largest mass gathering.
- Open Access
- 1 March 2023
- Published byĀ American Society of Tropical Medicine and HygieneĀ inĀ The American Journal of Tropical Medicine and Hygiene
- Vol.Ā 108 Ā (3) , 584-587
Clinico-Epidemiologic Characteristics of Patients Reported in the Mycotic Infections in COVID-19 Registry
We update results from the Mycotic Infections in COVID-19 (MUNCO) Registry, MayāSeptember 2021. Data collection from May to September 2021 yielded 728 cases from India, Nepal, Bangladesh, Thailand, and the United States. The cases consisted of mostly mucormycosis (97.6%), primarily rhinocerebral, and were analyzed to investigate clinical characteristics associated with negative outcomes. Patients were mostly diabetic (85%) and male (76%), with significant mortality (11.7%). All patients received treatment of coronavirus disease 2019 (COVID-19) as well as antifungal treatment. The crude mortality rate was 11.3% for mucormycosis and 22.7% formixed infections. This study demonstrates the utility of online databases in the collection of high-caliber data.
- 1 March 2023
- Published byĀ American Society of Tropical Medicine and HygieneĀ inĀ The American Journal of Tropical Medicine and Hygiene
- Vol.Ā 108 Ā (3) , 584-587
Open AccessClinico-Epidemiologic Characteristics of Patients Reported in the Mycotic Infections in COVID-19 Registry
We update results from the Mycotic Infections in COVID-19 (MUNCO) Registry, MayāSeptember 2021. Data collection from May to September 2021 yielded 728 cases from India, Nepal, Bangladesh, Thailand, and the United States. The cases consisted of mostly mucormycosis (97.6%), primarily rhinocerebral, and were analyzed to investigate clinical characteristics associated with negative outcomes. Patients were mostly diabetic (85%) and male (76%), with significant mortality (11.7%). All patients received treatment of coronavirus disease 2019 (COVID-19) as well as antifungal treatment. The crude mortality rate was 11.3% for mucormycosis and 22.7% formixed infections. This study demonstrates the utility of online databases in the collection of high-caliber data.
- Open Access
- 16 December 2022
- Published byĀ JMIR Publications Inc.Ā inĀ JMIR Preprints
Technology, Training, and Task Shifting at the Worldās Largest Mass Gathering in 2025: An Opportunity for Antibiotic Stewardship in India (Preprint)
UNSTRUCTURED The role of antibiotic overuse in intensifying selection pressures and contributing to the emergence of antimicrobial resistance is well established. The Kumbh Mela, a religious festival that occurs in 4 Indian cities of spiritual significance, is the worldās largest mass gathering, attracting over 80 million pilgrims in 2013. Digital syndromic surveillance from the 2013 and 2015 Melas demonstrated a consistent pattern of antibiotic overuse, with an antibiotic prescribing rate of up to 31% for all patient encounters. As preparations for the 2025 Kumbh Mela begin, task shifting, point-of-care diagnostic and digital tools, robust clinician training, and community awareness can promote the restrained and evidence-based use of antibiotics, minimizing the potential for the emergence of antimicrobial resistance at the worldās largest mass gathering.
- 16 December 2022
- Published byĀ JMIR Publications Inc.Ā inĀ JMIR Preprints
Open AccessTechnology, Training, and Task Shifting at the Worldās Largest Mass Gathering in 2025: An Opportunity for Antibiotic Stewardship in India (Preprint)
UNSTRUCTURED The role of antibiotic overuse in intensifying selection pressures and contributing to the emergence of antimicrobial resistance is well established. The Kumbh Mela, a religious festival that occurs in 4 Indian cities of spiritual significance, is the worldās largest mass gathering, attracting over 80 million pilgrims in 2013. Digital syndromic surveillance from the 2013 and 2015 Melas demonstrated a consistent pattern of antibiotic overuse, with an antibiotic prescribing rate of up to 31% for all patient encounters. As preparations for the 2025 Kumbh Mela begin, task shifting, point-of-care diagnostic and digital tools, robust clinician training, and community awareness can promote the restrained and evidence-based use of antibiotics, minimizing the potential for the emergence of antimicrobial resistance at the worldās largest mass gathering.
- Open Access
- 27 November 2022
- Published byĀ MDPI AGĀ inĀ Journal of Clinical Medicine
- Vol.Ā 11 Ā (23) , 7015
Factors Associated with Mortality in Coronavirus-Associated Mucormycosis: Results from Mycotic Infections in COVID-19 (MUNCO) Online Registry
Background: COVID-19-associated mucormycosis (CAM) is associated with high morbidity and mortality. MUNCO is an international database used to collect clinical data on cases of CAM in real time. Preliminary data from the Mycotic Infections in COVID-19 (MUNCO) online registry yielded 728 cases from May to September 2021 in four South Asian countries and the United States. A majority of the cases (694; 97.6%) consisted of a mucormycosis infection. The dataset allowed for the analysis of the risk factors for adverse outcomes from CAM and this analysis is presented in this paper. Methods: The submission of cases was aided by a direct solicitation and social media online. The primary endpoints were full recovery or death measured on day 42 of the diagnosis. All patients had histopathologically confirmed CAM. The groups were compared to determine the contribution of each patient characteristic to the outcome. Multivariable logistic regression models were used to model the probability of death after a CAM diagnosis. Results: The registry captured 694 cases of CAM. Within this, 341 could be analyzed as the study excluded patients with an unknown CAM recovery status due to either an interruption or a lack of follow up. The 341 viable cases consisted of 258 patients who survived after the completion of treatment and 83 patients who died during the period of observation. In a multivariable logistic regression model, the factors associated with an increased risk of mortality include old age (OR = 1.04, 95% CI 1.02ā1.07, p = 0.001), history of diabetes mellitus (OR 3.5, 95% CI 1.01ā11.9, p = 0.02) and a lower BMI (OR 0.9, 95% CI 0.82ā0.98, p = 0.03). Mucor localized to sinus disease was associated with 77% reduced odds of death (OR = 0.23, 95% CI 0.09ā0.57, p = 0.001), while cerebral mucor was associated with an increased odds of death (OR = 10.96, 95% CI 4.93ā24.36, p = ā¤0.0001). Conclusion: In patients with CAM, older age, a history of diabetes and a lower body mass index is associated with increased mortality. Disease limited to the sinuses without a cerebral extension is associated with a lower risk of mortality. Interestingly, the use of zinc and azithromycin were not associated with increased mortality in our study.
- 27 November 2022
- Published byĀ MDPI AGĀ inĀ Journal of Clinical Medicine
- Vol.Ā 11 Ā (23) , 7015
Open AccessFactors Associated with Mortality in Coronavirus-Associated Mucormycosis: Results from Mycotic Infections in COVID-19 (MUNCO) Online Registry
Background: COVID-19-associated mucormycosis (CAM) is associated with high morbidity and mortality. MUNCO is an international database used to collect clinical data on cases of CAM in real time. Preliminary data from the Mycotic Infections in COVID-19 (MUNCO) online registry yielded 728 cases from May to September 2021 in four South Asian countries and the United States. A majority of the cases (694; 97.6%) consisted of a mucormycosis infection. The dataset allowed for the analysis of the risk factors for adverse outcomes from CAM and this analysis is presented in this paper. Methods: The submission of cases was aided by a direct solicitation and social media online. The primary endpoints were full recovery or death measured on day 42 of the diagnosis. All patients had histopathologically confirmed CAM. The groups were compared to determine the contribution of each patient characteristic to the outcome. Multivariable logistic regression models were used to model the probability of death after a CAM diagnosis. Results: The registry captured 694 cases of CAM. Within this, 341 could be analyzed as the study excluded patients with an unknown CAM recovery status due to either an interruption or a lack of follow up. The 341 viable cases consisted of 258 patients who survived after the completion of treatment and 83 patients who died during the period of observation. In a multivariable logistic regression model, the factors associated with an increased risk of mortality include old age (OR = 1.04, 95% CI 1.02ā1.07, p = 0.001), history of diabetes mellitus (OR 3.5, 95% CI 1.01ā11.9, p = 0.02) and a lower BMI (OR 0.9, 95% CI 0.82ā0.98, p = 0.03). Mucor localized to sinus disease was associated with 77% reduced odds of death (OR = 0.23, 95% CI 0.09ā0.57, p = 0.001), while cerebral mucor was associated with an increased odds of death (OR = 10.96, 95% CI 4.93ā24.36, p = ā¤0.0001). Conclusion: In patients with CAM, older age, a history of diabetes and a lower body mass index is associated with increased mortality. Disease limited to the sinuses without a cerebral extension is associated with a lower risk of mortality. Interestingly, the use of zinc and azithromycin were not associated with increased mortality in our study.
- Open Access
- 9 September 2022
- Published byĀ MDPI AGĀ inĀ Journal of Clinical Medicine
- Vol.Ā 11 Ā (18) , 5312
COVID-19 Therapeutics: ImproviseāAdaptāLearn
āIn the midst of chaos, there is also opportunityāāSun Tzu, The Art of War
- 9 September 2022
- Published byĀ MDPI AGĀ inĀ Journal of Clinical Medicine
- Vol.Ā 11 Ā (18) , 5312
Open AccessCOVID-19 Therapeutics: ImproviseāAdaptāLearn
āIn the midst of chaos, there is also opportunityāāSun Tzu, The Art of War
- 9 August 2022
- Published byĀ BMJĀ inĀ BMJ Case Reports
- Vol.Ā 15 Ā (8) , e247893
Quantitative B and T cell abnormalities in four patients presenting with mucormycosis and prior asymptomatic COVID-19 infection
India saw an unprecedented and rapid rise of invasive coronavirus-associated mucormycosis (CAM) during the delta COVID-19 surge. There is little known about the pathophysiology and if there is a direct causation between the COVID-19 infection and invasive CAM. While the traditional risk factors such as uncontrolled diabetes and other immunocompromising conditions are recognised, there could be several COVID-19-induced phenomena that may predispose the patients to develop CAM and are yet unrecognised. It has been proposed that prior severe COVID-19 is associated with invasive CAM. This could be due to the increased use of immunomodulators or the direct effects of the COVID-19 infection. We report four patients with CAM during the delta surge who did not have prior known COVID-19 infection but on subsequent testing had positive antibodies suggesting past asymptomatic infection. We report the quantitative abnormalities in lymphocyte subsets in all four patients and report CD19+ B cell lymphopenia and reduced percentage of CD27+ CD45RA+ naĆÆve helper T cells. CAM may occur in patients after asymptomatic COVID-19 infection, in the absence of systemic corticosteroid and immunomodulator use, and may reflect underlying immune abnormalities possibly attributable to or unmasked by prior COVID-19 infection.
- 9 August 2022
- Published byĀ BMJĀ inĀ BMJ Case Reports
- Vol.Ā 15 Ā (8) , e247893
Quantitative B and T cell abnormalities in four patients presenting with mucormycosis and prior asymptomatic COVID-19 infection
India saw an unprecedented and rapid rise of invasive coronavirus-associated mucormycosis (CAM) during the delta COVID-19 surge. There is little known about the pathophysiology and if there is a direct causation between the COVID-19 infection and invasive CAM. While the traditional risk factors such as uncontrolled diabetes and other immunocompromising conditions are recognised, there could be several COVID-19-induced phenomena that may predispose the patients to develop CAM and are yet unrecognised. It has been proposed that prior severe COVID-19 is associated with invasive CAM. This could be due to the increased use of immunomodulators or the direct effects of the COVID-19 infection. We report four patients with CAM during the delta surge who did not have prior known COVID-19 infection but on subsequent testing had positive antibodies suggesting past asymptomatic infection. We report the quantitative abnormalities in lymphocyte subsets in all four patients and report CD19+ B cell lymphopenia and reduced percentage of CD27+ CD45RA+ naĆÆve helper T cells. CAM may occur in patients after asymptomatic COVID-19 infection, in the absence of systemic corticosteroid and immunomodulator use, and may reflect underlying immune abnormalities possibly attributable to or unmasked by prior COVID-19 infection.