About

Sanjiv Devendra Mehta, MD, MSCE, MBE, is an Assistant Professor of Anesthesiology and Critical Care at the Children's Hospital of Philadelphia. He is also a member of the Department of Biomedical and Health Informatics and a core faculty member at the Clinical Futures research institute within the Children's Hospital of Philadelphia. His educational background includes a BA in Economics-Philosophy and Biology from Columbia University, an MD from the University of Pennsylvania Perelman School of Medicine, an MBE in Bioethics from the University of Pennsylvania, and an MSCE in Clinical Epidemiology expected in 2025. His research focuses on critical care medicine, with particular interests in machine learning-based prediction of critical deterioration in pediatric intensive care units, the use of AI chatbots to support parental understanding in the PICU, and the analysis of neighborhood factors and socioeconomic disparities in pediatric ICU transfers. He has contributed to studies on emergency transfers, prearrest vital sign abnormalities, and prognostic models in intensive care, emphasizing the application of digital health tools and data-driven approaches to improve pediatric critical care outcomes.

Research topics

• Medicine
• Emergency medicine
• Intensive care medicine
• Medical emergency
• Pediatrics

Selected publications

• Feasibility of a Large Language Model Chatbot to Support Parental Understanding in the PICU

  Critical Care Explorations · 2026-03-20

  articleOpen accessSenior author

  OBJECTIVES: To evaluate the feasibility of a large language model (LLM)-based chatbot for answering parental questions in the PICU and inform design of a randomized controlled trial (RCT). DESIGN: Prospective single-arm feasibility study conducted from August 2024 to December 2024. SETTING: Quaternary PICU. SUBJECTS: Fourteen parents of children admitted to the PICU. INTERVENTIONS: Parents engaged in 10-minute sessions with a HIPAA-compliant GPT-4o- (Generative Pretrained Transformer 4o, OpenAI, San Francisco, CA) based chatbot prompted with patient-specific electronic health record (EHR) data. MEASUREMENTS AND MAIN RESULTS: Feasibility was assessed through four criteria: parental engagement and satisfaction, provider perceptions, accuracy and safety, and recruitment. Of 16 eligible parents, 14 enrolled and completed all procedures (87.5% recruitment rate). Parents asked a median of six questions (range, 3-13) with 96% positive real-time satisfaction ratings. Post-interaction surveys demonstrated high perceived value (median, 5.0/6.0 across all domains; Net Promoter Score [NPS] +57). Of 1225 chatbot-generated sentences evaluated, 99.3% were accurate with all eight errors classified as minor (inter-rater reliability: Gwet's AC2, a chance-corrected inter-rater agreement coefficient, = 0.98; 95% CI, 0.97-0.99). Healthcare providers rated response quality highly (median, 5.0/6.0), although physicians expressed greater comfort with bedside use of the tool than nurses (5.0 vs. 4.0; p = 0.004). Sample size calculations using NPS as the primary endpoint suggest enrolling 135 participants would provide adequate power for a future RCT. CONCLUSIONS: An EHR-informed LLM chatbot demonstrated high parental engagement and satisfaction, positive provider perception, and high accuracy and safety, supporting progression to a RCT.

  PublisherDOI

• Machine Learning-Based Prediction of Critical Deterioration in the PICU

  Critical Care Explorations · 2026-04-01

  articleOpen access1st authorCorresponding

  BACKGROUND: Existing PICU early warning systems lack sufficient accuracy and timeliness for effective preparation. Machine learning approaches may improve prediction of critical deterioration events (CDEs), but their operational utility relative to existing tools remains unclear. OBJECTIVES: To develop a machine learning model for early detection of CDEs and evaluate operational utility against existing tools using a novel alert burden analysis. DERIVATION COHORT: PICU admissions (ages 0-24 yr, stay ≥ 24 hr) at a quaternary children's hospital from 2014 to 2020 (n = 12,771 patients; 21,141 admissions). CDEs (6% of patients) included cardiopulmonary resuscitation, extracorporeal membrane oxygenation initiation, dilute epinephrine administration, or unplanned intubation. VALIDATION COHORT: Temporally distinct PICU admissions from 2021 to 2022 (n = 5144 patients; 6929 admissions; 6% CDE rate). PREDICTION MODEL: An ensemble of extreme gradient-boosted models (PICU Warning INdex [P-WIN]) trained to predict CDEs at 1-12-hour horizons using 550 features derived from demographics, medications, laboratory results, and vital signs. RESULTS: P-WIN demonstrated excellent discrimination at 2-hour (area under the receiver operating characteristic curve [AUROC], 0.95 [95% CI, 0.94-0.96] and area under the precision-recall curve [AUPRC], 0.76 [95% CI, 0.72-0.80]) and 12-hour horizons (AUROC, 0.93 [95% CI, 0.92-0.94] and AUPRC, 0.68 [95% CI, 0.64-0.72]). To alert before 80% of events, P-WIN generated 0.20 alerts per patient-day at a median 10.17 hours before CDE. Compared with the existing rule-based PICU Warning Tool (alerting before 38% of events), P-WIN generated one-third the alert burden at equivalent sensitivity (0.03 vs. 0.10 alerts per patient-day). CONCLUSIONS: P-WIN accurately predicted PICU CDEs up to 12 hours in advance with low alert burden, providing a viable opportunity for shifting care from reactive rescue to proactive, resource-intensive preparation and prevention.

  PublisherDOI

• 406: AI CHATBOTS IN THE PICU: PARENTAL ENTHUSIASM CONTRASTS WITH SOCIOECONOMIC USAGE DISPARITIES

  Critical Care Medicine · 2025-01-01

  article
  PublisherDOI

• Prearrest vital sign abnormalities are associated with adverse outcomes in pediatric ICU cardiac arrest: a get with the guidelines-resuscitation analysis

  Resuscitation · 2025-09-25

  articleOpen access1st authorCorresponding

  AIM: We aimed to quantitatively describe vital sign abnormalities prior to pediatric IHCA and evaluate whether the severity of abnormalities was independently associated with survival. METHODS: In a retrospective cohort study using the American Heart Association's Get with The Guidelines-Resuscitation® registry, children with ≥1 min of cardiopulmonary resuscitation (CPR) in an Intensive Care Unit (ICU) from 2007 to 2022 with prearrest vital signs were included. Vital signs most proximate to CPR (10-120 min prior) were classified as abnormal (HR or RR >95th, SBP or DBP <5th percentile for age). Multivariable regression adjusted for age, illness category, prearrest conditions, and prearrest interventions assessed the associations between vital sign abnormalities and outcomes (primary: survival to hospital discharge, secondary: return of spontaneous circulation [ROSC]). RESULTS: Of 2875 IHCA patients meeting inclusion criteria, 1790 (62.3 %) had at least one abnormal vital sign. Patients with vital sign abnormalities were older, had non-surgical illness categories, and higher prevalence of prearrest illnesses and interventions. Low SBP (<5%) was the vital sign with the lowest odds of survival to hospital discharge (aOR 0.56 [95 %CI 0.46-0.68], p < 0.01) and ROSC (aOR 0.63 [95 %CI 0.54-0.73], p < 0.01). There was a stepwise decrease in the adjusted odds of survival for each additional abnormal vital sign (1 vs 0: aOR 0.62 [95 %CI 0.51-0.76], p < 0.01; 2 vs 1: 0.72 [95 %CI 0.53-0.97] p = 0.03; 3 vs 2: 0.53 [95 %CI 0.33-0.86] p < 0.01). CONCLUSIONS: Prearrest vital sign abnormalities are common in pediatric ICU IHCA and independently associated with worse outcomes, emphasizing the need for prompt detection and intervention to improve outcomes.

  PublisherDOI

• 1536: DEVELOPMENT AND VALIDATION OF A PEDIATRIC ICU DETERIORATION PREDICTION MODEL USING MACHINE LEARNING

  Critical Care Medicine · 2025-01-01

  article1st authorCorresponding
  PublisherDOI

• 383: DEVELOPMENT AND VALIDATION OF A PREDICTIVE MODEL FOR NON-ESCALATION IN BRONCHIOLITIS

  Critical Care Medicine · 2025-01-01

  articleSenior author
  PublisherDOI

• Early exposure to hemodynamic point-of-care ultrasound during pediatric septic shock resuscitation is associated with decreased fluid overload

  Intensive Care Medicine – Paediatric and Neonatal · 2025-09-24

  articleOpen access

  Abstract Objective Fluid overload (FO) after pediatric septic shock resuscitation increases the risk of secondary organ failure and long-term morbidity. We hypothesized that early hemodynamic point-of-care ultrasound (hPOCUS) use in pediatric septic shock would be associated with decreased FO. Methods Retrospective, observational study between 2015 and 2018 in a large academic Pediatric Intensive Care Unit (PICU) of children &lt; 18 years receiving ≥ 40 mL/kg of bolus fluids or vasoactive infusion for treatment of septic shock. %FO and severe FO (%FO ≥ 15%) at 72 h after shock onset were compared between children with hPOCUS exposure &lt; 6 h after shock onset and those without. %FO was calculated: [(fluid in – out)/weight on PICU admission]*100. The association between hPOCUS &lt; 6 h after septic shock onset and %FO and severe FO were evaluated. Results Of 591 children included, 115 (19.4%) had hPOCUS within 6 h. Children with early hPOCUS vs. without had higher PIM-3; median 4.3, IQR 2.9–9.4 vs. 3.2, IQR 0.9–5.7, p = 0.0012. %FO was not significantly different in the early hPOCUS group, median difference -1.6%, 95% CI -3.2 to 0.03, p = 0.06. After controlling for confounders, %FO was significantly lower in the early hPOCUS group, ß-coefficient= -2.76, 95% CI -4.7 to -0.6, p = 0.012. Those with early hPOCUS had lower occurrence of severe FO 11.3% vs. 22.9%, p = 0.006; adjusted OR 0.41, 95% CI 0.22 to 0.76, p = 0.005. Conclusions Early hPOCUS assessment during septic shock resuscitation was independently associated with decreased FO. Prospective research is needed to optimize hPOCUS use in pediatric septic shock management.

  PublisherOA PDFDOI

• Hypothesis and error types in studies

  Elsevier eBooks · 2025-01-01

  book-chapter1st authorCorresponding
  PublisherDOI

• An Unusual Case of TH Mutation Presenting with Titubation: Movement Disorder

  Journal of Academy of Pediatric Neurology · 2025-11-14

  articleOpen access

  Introduction: Tyrosine hydroxylase (TH) is a key enzyme in catecholamine biosynthesis, playing a crucial role in dopamine (DOPA) production.Tyrosine hydroxylase deficiency has a broad clinical spectrum ranging from DOPA-responsive dystonia to severe infantile encephalopathy.Although classically presenting with hypokinesia, rigidity, and dystonia, atypical phenotypes can mimic other neurological disorders, often leading to misdiagnosis and delayed treatment.Case description: In a family of four, the patient's elder sister having an uneventful birth, had global developmental delay and presented as a cerebral palsy (CP) mimic, initial investigations and magnetic resonance imaging (MRI) Brain were normal, on suspicion of genetic abnormality, whole exome sequencing was done showing homozygous mutation in TH gene and DOPA trial was started, following which there is marked improvement.The patient, a 6-month-old male, had an uneventful birth, was neurodevelopmentally normal till 5 months of age, when parents noticed head titubation.Neurological examination revealed rhythmic oscillatory movements without signs of spasticity, dystonia, or developmental regression.more in the morning and increased crying, these movements also improved with a low dose of DOPA.Discussion: This case highlights an unusual presentation of TH deficiency with isolated titubation as an early sign.It underscores the expanding phenotypic variability of the disorder and the potential for misdiagnosis, particularly when the presentation deviates from classical motor findings.Early recognition and treatment with DOPA replacement can result in significant clinical recovery.Moreover, the presence of a similarly affected sibling earlier diagnosed as CP mimics emphasizes the need for heightened clinical suspicion and appropriate workup in familial cases of unexplained movement disorders.

  PublisherDOI

• Temporary Femoral Central Venous Catheters in the PICU: Two-Center, Retrospective Cohort Study of Catheter Tip Position and Symptomatic Venous Thromboembolism, 2016–2021

  Pediatric Critical Care Medicine · 2025-02-11 · 2 citations

  articleOpen access

  OBJECTIVES: We aimed to determine the frequency and variables associated with low femoral central venous catheter (fCVC) tip position. We also examined the association between tip position and symptomatic venous thromboembolism (VTE). DESIGN: Retrospective cohort from two PICUs. SETTING: Quaternary academic children's hospitals, 2016-2021. PATIENTS: Children (age <18 yr) in the PICU who underwent temporary fCVC placement. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: Low fCVC tip position occurs when the tip is inferior to the fifth lumbar vertebra (L5) on a postprocedural abdominal radiograph. Of 936 patients: 56.3% were 1-12 years old, and 80.0% had normal weight-for-age z score. fCVC tip position was low in 67.3% of patients. In the multivariable model, older age, earlier years of placement, and higher weight-for-age were associated with low fCVC tip position. Symptomatic fCVC-associated VTE occurred in 8.8% of patients, with a rate of 16.5 per 1000 CVC days (interquartile range, 13.1-20.5 per 1000 CVC days). The percentage of VTE in low vs. recommended fCVC tip position and VTE (8.6% vs. 9.2%) were equivalent (two one-sided z-tests; p < 0.001). Furthermore, in the multivariable model, we failed to identify an association between low fCVC tip position, relative to the recommended tip position, and greater odds of VTE (OR, 1.58 [95% CI, 0.92-2.69). However, we cannot exclude the possibility of low fCVC tip position being associated with up to 2.6-fold greater odds of symptomatic VTE. CONCLUSIONS: In our two PICUs, 2016-2021, low fCVC tip position occurred in two-thirds of placements and was associated with older age and higher weight-for-age patients. fCVC-associated VTE occurred in one-in-11-catheter placements, with the raw percentage of fCVCs and subsequent VTE in low and recommended tip position being equivalent. However, the multivariable modeling indicates that future research into the relationship between tip position and VTE requires ongoing surveillance and work.

  PublisherOA PDFDOI

Frequent coauthors

• Sameer Nagda

  82 shared

• R. Sean Churchill

  81 shared

• Brent B. Wiesel

  81 shared

• Derek J. Donegan

  University of Pennsylvania

  69 shared

• Tiffany Tedore

  Presbyterian Hospital

  42 shared

• Robert Zura

  Hamilton Health Sciences

  41 shared

• Gregory J. Della Rocca

  University of Maryland, Baltimore

  37 shared

• Robert A. Hymes

  Inova Fairfax Hospital

  36 shared

Education

• Medical Degree

  Perelman School of Medicine, University of Pennsylvania

  2016