Joshua Broder
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
Research Subjects
Joshua Broder has a total of 130 co-authors. Here are the top co-authors by number of shared publications:
- Christopher R. Carpenter
Emergency Medicine, Mayo Clinic, Rochester, Minnesota, USA
5 shared publications - M. Fernanda Bellolio
Department of Emergency Medicine, Mayo Clinic, Rochester, Minnesota, USA
5 shared publications - Carl D. Herickhoff
Department of Radiology, Stanford University School of Medicine, Palo Alto, CA, 94304 USA
4 shared publications - Lucas Oliveira Junqueira e Silva
Universidade Federal do Rio Grande do Sul
4 shared publications - Matthew R. Morgan
Department of Biomedical Engineering, Duke University, Durham, NC, USA
4 shared publications - Andrew Charles Meltzer
The George Washington University School of Medicine and Health Sciences, Department of Emergency Medicine, Washington, DC, USA
3 shared publications
Latest Publications
- Open Access
- 15 October 2026
- Published by Frontiers Media SA in Frontiers in Medical Technology
- Vol. 8, 1955568
Automatic control of arterial blood pressure: computer, benchtop, and animal models
Background: Timely, accurate, and stable control of blood pressure (BP) in critical illness is essential for good clinical outcomes. The current clinical standard entails human operators intervening between the monitor and infusion pump, introducing delay, potential error, and risk of over- and undershoot of therapeutic goals. We developed and tested a closed-loop, automatic control system for BP using computer, benchtop, and animal models. Methods: All models employed a proportional-integral-derivative (PID) algorithm to control mean arterial pressure (MAP). The three components are summed to determine the final output, corresponding to the infusion rate of fluids or vasoactive drugs. The computer model simulated an automatically controlled drug infusion to regulate MAP. The benchtop model added physical components (clinical cardiac monitor, microcontroller, medication pump, and pressure transducer), adjusting the hydrostatic pressure in a closed system to meet a target. In the animal model, the MAP of anesthetized swine undergoing surgery was regulated using an automatically controlled norepinephrine or clevidipine infusion. For the animal model, statistical inference and comparison between trials could not be made given the sequential, non-independent design; quantities are descriptive only, not inferential. We applied common control system metrics including rise time, settling time, and on-target percentage (OTP). Results: The computer model rapidly reached and maintained target MAP ± 5 mmHg with OTP of 90% (95% CI 89.4%, 90.6%) at 240 simulated seconds (ss) and 95% (95% CI 94.6%, 95.4%) at 294 ss. The benchtop system reached and maintained target MAP with a mean rise time of 0.47 ± SD 0.15 min, mean settling time of 0.62 ± SD 0.18 min, and mean OTP (using a tolerance of MAP ± 5 mmHg) of 91.25 ± 9.68%. In a swine model, automatic norepinephrine augmented MAP by > 15 mmHg with mean rise time of 2.25 (SD 0.65) min and mean OTP of 56.9% (SD 22.7%). No trial maintained an equilibrium value of target MAP ± 5 mmHg; therefore, no settling time could be calculated. Clevidipine decreased MAP by >15 mmHg with mean rise time of 1.00 (SD 0.40) min, mean settling time of 2.19 (SD 1.45) min, and mean OTP of 93.8% (SD 7.18%). The results of the swine model were confounded by concurrent non-survival surgery. Within this limitation, the swine model demonstrated that the automatic control system was technically feasible. Conclusions: These proof-of-concept studies utilizing computer, benchtop, and animal models provide initial evidence that an automatic PID algorithm has potential to control BP across a range of starting BP. Future studies can optimize the system and compare its performance with human operators in advance of potential human clinical trials.
- 15 October 2026
- Published by Frontiers Media SA in Frontiers in Medical Technology
- Vol. 8, 1955568
Open AccessAutomatic control of arterial blood pressure: computer, benchtop, and animal models
Background: Timely, accurate, and stable control of blood pressure (BP) in critical illness is essential for good clinical outcomes. The current clinical standard entails human operators intervening between the monitor and infusion pump, introducing delay, potential error, and risk of over- and undershoot of therapeutic goals. We developed and tested a closed-loop, automatic control system for BP using computer, benchtop, and animal models. Methods: All models employed a proportional-integral-derivative (PID) algorithm to control mean arterial pressure (MAP). The three components are summed to determine the final output, corresponding to the infusion rate of fluids or vasoactive drugs. The computer model simulated an automatically controlled drug infusion to regulate MAP. The benchtop model added physical components (clinical cardiac monitor, microcontroller, medication pump, and pressure transducer), adjusting the hydrostatic pressure in a closed system to meet a target. In the animal model, the MAP of anesthetized swine undergoing surgery was regulated using an automatically controlled norepinephrine or clevidipine infusion. For the animal model, statistical inference and comparison between trials could not be made given the sequential, non-independent design; quantities are descriptive only, not inferential. We applied common control system metrics including rise time, settling time, and on-target percentage (OTP). Results: The computer model rapidly reached and maintained target MAP ± 5 mmHg with OTP of 90% (95% CI 89.4%, 90.6%) at 240 simulated seconds (ss) and 95% (95% CI 94.6%, 95.4%) at 294 ss. The benchtop system reached and maintained target MAP with a mean rise time of 0.47 ± SD 0.15 min, mean settling time of 0.62 ± SD 0.18 min, and mean OTP (using a tolerance of MAP ± 5 mmHg) of 91.25 ± 9.68%. In a swine model, automatic norepinephrine augmented MAP by > 15 mmHg with mean rise time of 2.25 (SD 0.65) min and mean OTP of 56.9% (SD 22.7%). No trial maintained an equilibrium value of target MAP ± 5 mmHg; therefore, no settling time could be calculated. Clevidipine decreased MAP by >15 mmHg with mean rise time of 1.00 (SD 0.40) min, mean settling time of 2.19 (SD 1.45) min, and mean OTP of 93.8% (SD 7.18%). The results of the swine model were confounded by concurrent non-survival surgery. Within this limitation, the swine model demonstrated that the automatic control system was technically feasible. Conclusions: These proof-of-concept studies utilizing computer, benchtop, and animal models provide initial evidence that an automatic PID algorithm has potential to control BP across a range of starting BP. Future studies can optimize the system and compare its performance with human operators in advance of potential human clinical trials.
- Open Access
- 1 February 2025
- Published by Elsevier in JEM Reports
- Vol. 4 (1)
3D-augmentation of 2D ultrasound for appendicitis diagnosis: A cross-sectional pilot study
- 1 February 2025
- Published by Elsevier in JEM Reports
- Vol. 4 (1)
Open Access3D-augmentation of 2D ultrasound for appendicitis diagnosis: A cross-sectional pilot study
- Open Access
- 21 November 2024
- Published by California Digital Library (CDL) in Western Journal of Emergency Medicine
- Vol. 26 (1) , 161-170
A Machine Learning Algorithm to Predict Medical Device Recall by the Food and Drug Administration
Introduction: Medical device recalls are important to the practice of emergency medicine, as unsafe devices include many ubiquitous items in emergency care, such as vascular access devices, ventilators, infusion pumps, video laryngoscopes, pulse oximetry sensors, and implantable cardioverter defibrillators. Identification of dangerous medical devices as early as possible is necessary to minimize patient harms while avoiding false positives to prevent removal of safe devices from use. While the United States Food and Drug Administration (FDA) employs an adverse event reporting program (MedWatch) and database (MAUDE), other data sources and methods might have utility to identify potentially dangerous medical devices. Our objective was to evaluate the sensitivity, specificity, and accuracy of a machine learning (ML) algorithm using publicly available data to predict medical device recalls by the FDA. Methods: We identified recalled medical devices (RMD) and non-recalled medical devices (NRMD) using the FDA’s website and online database. We constructed an ML algorithm (random forest regressor) that automatically searched Google Trends and PubMed for the RMDs and NRMDs. The algorithm was trained using 400 randomly selected devices and then tested using 100 unique random devices. The algorithm output a continuous value (0–1) for recall probability for each device, which were rounded for dichotomous analysis. We determined sensitivity, specificity, and accuracy for each of three time periods prior to recall (T-3, 6, or 12 months), using FDA recall status as the reference standard. The study adhered to relevant items of the Standards for Reporting Diagnostic accuracy studies (STARD) guidelines. Results: Using a rounding threshold of 0.5, sensitivities for T-3, T-6, and T-12 were 89% (95%confidence interval [CI] 69–97), 90% (95% CI 70–97), and 75% (95% CI 53–89). Specificity was 100%(95% CI 95–100) for all three time periods. Accuracy was 98% (95% CI 93–99) for T-3 and T-6, and 95%(95% CI 89–99) for T-12. Using tailored thresholds yielded similar results. Conclusion: An ML algorithm accurately predicted medical device recall status by the FDA with lead times as great as 12 months. Future research could incorporate longer lead times and data sources including FDA reports and prospectively test the ability of ML algorithms to predict FDA recall.
- 21 November 2024
- Published by California Digital Library (CDL) in Western Journal of Emergency Medicine
- Vol. 26 (1) , 161-170
Open AccessA Machine Learning Algorithm to Predict Medical Device Recall by the Food and Drug Administration
Introduction: Medical device recalls are important to the practice of emergency medicine, as unsafe devices include many ubiquitous items in emergency care, such as vascular access devices, ventilators, infusion pumps, video laryngoscopes, pulse oximetry sensors, and implantable cardioverter defibrillators. Identification of dangerous medical devices as early as possible is necessary to minimize patient harms while avoiding false positives to prevent removal of safe devices from use. While the United States Food and Drug Administration (FDA) employs an adverse event reporting program (MedWatch) and database (MAUDE), other data sources and methods might have utility to identify potentially dangerous medical devices. Our objective was to evaluate the sensitivity, specificity, and accuracy of a machine learning (ML) algorithm using publicly available data to predict medical device recalls by the FDA. Methods: We identified recalled medical devices (RMD) and non-recalled medical devices (NRMD) using the FDA’s website and online database. We constructed an ML algorithm (random forest regressor) that automatically searched Google Trends and PubMed for the RMDs and NRMDs. The algorithm was trained using 400 randomly selected devices and then tested using 100 unique random devices. The algorithm output a continuous value (0–1) for recall probability for each device, which were rounded for dichotomous analysis. We determined sensitivity, specificity, and accuracy for each of three time periods prior to recall (T-3, 6, or 12 months), using FDA recall status as the reference standard. The study adhered to relevant items of the Standards for Reporting Diagnostic accuracy studies (STARD) guidelines. Results: Using a rounding threshold of 0.5, sensitivities for T-3, T-6, and T-12 were 89% (95%confidence interval [CI] 69–97), 90% (95% CI 70–97), and 75% (95% CI 53–89). Specificity was 100%(95% CI 95–100) for all three time periods. Accuracy was 98% (95% CI 93–99) for T-3 and T-6, and 95%(95% CI 89–99) for T-12. Using tailored thresholds yielded similar results. Conclusion: An ML algorithm accurately predicted medical device recall status by the FDA with lead times as great as 12 months. Future research could incorporate longer lead times and data sources including FDA reports and prospectively test the ability of ML algorithms to predict FDA recall.
- 1 November 2024
- Published by Wiley in Academic Emergency Medicine
- Vol. 32 (2) , 179-182
Miles to go before we sleep: Does increasing abdominal computed tomography utilization really improve patient‐oriented outcomes?
- 1 November 2024
- Published by Wiley in Academic Emergency Medicine
- Vol. 32 (2) , 179-182
Miles to go before we sleep: Does increasing abdominal computed tomography utilization really improve patient‐oriented outcomes?
- 23 November 2023
- Published by Elsevier in Academic Radiology
- Vol. 31 (2) , 536-537
Predicting the Aggressiveness of Papillary Thyroid Carcinoma Preoperatively: Early Steps in the Right Direction
- 23 November 2023
- Published by Elsevier in Academic Radiology
- Vol. 31 (2) , 536-537
Predicting the Aggressiveness of Papillary Thyroid Carcinoma Preoperatively: Early Steps in the Right Direction
- 30 May 2023
- Published by Wiley in Academic Emergency Medicine
- Vol. 30 (10) , 1073-1074
Easy as Pi? Why geometry might save us from tough percutaneous catheter insertions
- 30 May 2023
- Published by Wiley in Academic Emergency Medicine
- Vol. 30 (10) , 1073-1074
Easy as Pi? Why geometry might save us from tough percutaneous catheter insertions
- 11 August 2022
- Published by Wiley in Academic Emergency Medicine
- Vol. 29 (9) , 1148-1150
Reply to “Letter to the Editor, re: GRACE ‐2: Low‐Risk , Recurrent Abdominal Pain in the Emergency Department”
- 11 August 2022
- Published by Wiley in Academic Emergency Medicine
- Vol. 29 (9) , 1148-1150
Reply to “Letter to the Editor, re: GRACE ‐2: Low‐Risk , Recurrent Abdominal Pain in the Emergency Department”
- 11 May 2022
- Published by Wiley in Academic Emergency Medicine
- Vol. 29 (5) , 526-560
Guidelines for Reasonable and Appropriate Care in the Emergency Department 2 (GRACE‐2): Low‐risk, recurrent abdominal pain in the emergency department
This second Guideline for Reasonable and Appropriate Care in the Emergency Department (GRACE‐2) from the Society for Academic Emergency Medicine is on the topic “low‐risk, recurrent abdominal pain in the emergency department.” The multidisciplinary guideline panel applied the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach to assess the certainty of evidence and strength of recommendations regarding four priority questions for adult emergency department patients with low‐risk, recurrent, undifferentiated abdominal pain. The intended population includes adults with multiple similar presentations of abdominal signs and symptoms recurring over a period of months or years. The panel reached the following recommendations: (1) if a prior negative computed tomography of the abdomen and pelvis (CTAP) has been performed within 12 months, there is insufficient evidence to accurately identify populations in whom repeat CTAP imaging can be safely avoided or routinely recommended; (2) if CTAP with IV contrast is negative, we suggest against ultrasound unless there is concern for pelvic or biliary pathology; (3) we suggest that screening for depression and/or anxiety may be performed during the ED evaluation; and (4) we suggest an opioid‐minimizing strategy for pain control.EXECUTIVE SUMMARY: The GRACE‐2 writing group developed clinically relevant questions to address the care of adult patients with low‐risk, recurrent, previously undifferentiated abdominal pain in the emergency department (ED). Four patient‐intervention‐comparison‐outcome‐time (PICOT) questions were developed by consensus of the writing group, who performed a systematic review of the literature and then synthesized direct and indirect evidence to formulate recommendations, following GRADE methodology. The writing group found that despite the commonality and relevance of these questions in emergency care, the quantity and quality of evidence were very limited, and even fundamental definitions of the population and outcomes of interest are lacking. Future research opportunities include developing precise and clinically relevant definitions of low‐risk, recurrent, undifferentiated abdominal pain and determining the scope of the existing populations in terms of annual national ED visits for this complaint, costs of care, and patient and provider preferences.
- 11 May 2022
- Published by Wiley in Academic Emergency Medicine
- Vol. 29 (5) , 526-560
Guidelines for Reasonable and Appropriate Care in the Emergency Department 2 (GRACE‐2): Low‐risk, recurrent abdominal pain in the emergency department
This second Guideline for Reasonable and Appropriate Care in the Emergency Department (GRACE‐2) from the Society for Academic Emergency Medicine is on the topic “low‐risk, recurrent abdominal pain in the emergency department.” The multidisciplinary guideline panel applied the Grading of Recommendations Assessment, Development and Evaluation (GRADE) approach to assess the certainty of evidence and strength of recommendations regarding four priority questions for adult emergency department patients with low‐risk, recurrent, undifferentiated abdominal pain. The intended population includes adults with multiple similar presentations of abdominal signs and symptoms recurring over a period of months or years. The panel reached the following recommendations: (1) if a prior negative computed tomography of the abdomen and pelvis (CTAP) has been performed within 12 months, there is insufficient evidence to accurately identify populations in whom repeat CTAP imaging can be safely avoided or routinely recommended; (2) if CTAP with IV contrast is negative, we suggest against ultrasound unless there is concern for pelvic or biliary pathology; (3) we suggest that screening for depression and/or anxiety may be performed during the ED evaluation; and (4) we suggest an opioid‐minimizing strategy for pain control.EXECUTIVE SUMMARY: The GRACE‐2 writing group developed clinically relevant questions to address the care of adult patients with low‐risk, recurrent, previously undifferentiated abdominal pain in the emergency department (ED). Four patient‐intervention‐comparison‐outcome‐time (PICOT) questions were developed by consensus of the writing group, who performed a systematic review of the literature and then synthesized direct and indirect evidence to formulate recommendations, following GRADE methodology. The writing group found that despite the commonality and relevance of these questions in emergency care, the quantity and quality of evidence were very limited, and even fundamental definitions of the population and outcomes of interest are lacking. Future research opportunities include developing precise and clinically relevant definitions of low‐risk, recurrent, undifferentiated abdominal pain and determining the scope of the existing populations in terms of annual national ED visits for this complaint, costs of care, and patient and provider preferences.
- 23 April 2022
- Published by Wiley in Academic Emergency Medicine
- Vol. 29 (5) , 674-677
A candle in the dark: The role of indirect evidence in emergency medicine clinical practice guidelines
- 23 April 2022
- Published by Wiley in Academic Emergency Medicine
- Vol. 29 (5) , 674-677
A candle in the dark: The role of indirect evidence in emergency medicine clinical practice guidelines
- 13 December 2021
- Published by Wiley in Academic Emergency Medicine
- Vol. 29 (5) , 630-648
Repeat computed tomography in recurrent abdominal pain: An evidence synthesis for guidelines for reasonable and appropriate care in the emergency department
Objective Computed tomography (CT) imaging is frequently obtained for recurrent abdominal pain after a prior emergency department (ED) evaluation. We evaluate the utility of repeat CT imaging following an indeterminate index CT in low-risk abdominal pain adult ED patients. Methods An electronic search was designed for the Patient-Intervention-Control-Outcome-Timing (PICOT) question: (P) adult patients with low- risk, recurrent and previously undifferentiated atraumatic abdominal pain presenting to the ED after an index negative CT within 12-months, (I) repeat CT versus (C) no repeat CT, for (O) abdominal surgery or other invasive procedure, mortality, identification of potentially life-threatening diagnosis, hospital and ICU admission rates, return ED visit (T) all within 30-days. Four reviewers independently selected evidence for inclusion and then synthesized the results around the most prevalent themes of repeat CT timing, diagnostic yield, ionizing radiation exposure, and predictors of repetitive imaging. Results Although 637 manuscripts and abstracts were identified, no direct evidence was found. Thirteen documents were synthesized as indirect evidence. None of the indirect evidence defined a low-risk subset of abdominal pain, nor did investigators describe whether re-imaging occurred for complaints similar to the initial ED evaluation. Included studies did not describe the index CT findings and some reported explanatory findings noted on the original CT for which repeat CTs might have been indicated. The timeframe for a repeat CT ranged from hours to one year. The frequency of repeat CTs (2-47%) varied across studies as did the yield of imaging to alter downstream clinical decision-making (range 5-67%). Conclusion Due to the absence of direct evidence our scoping review is unable to provide high-quality evidence-based recommendations upon which to confidently base an imaging practice guideline. There is no evidence to support or refute performing a CT for low-risk recurrent abdominal pain.
- 13 December 2021
- Published by Wiley in Academic Emergency Medicine
- Vol. 29 (5) , 630-648
Repeat computed tomography in recurrent abdominal pain: An evidence synthesis for guidelines for reasonable and appropriate care in the emergency department
Objective Computed tomography (CT) imaging is frequently obtained for recurrent abdominal pain after a prior emergency department (ED) evaluation. We evaluate the utility of repeat CT imaging following an indeterminate index CT in low-risk abdominal pain adult ED patients. Methods An electronic search was designed for the Patient-Intervention-Control-Outcome-Timing (PICOT) question: (P) adult patients with low- risk, recurrent and previously undifferentiated atraumatic abdominal pain presenting to the ED after an index negative CT within 12-months, (I) repeat CT versus (C) no repeat CT, for (O) abdominal surgery or other invasive procedure, mortality, identification of potentially life-threatening diagnosis, hospital and ICU admission rates, return ED visit (T) all within 30-days. Four reviewers independently selected evidence for inclusion and then synthesized the results around the most prevalent themes of repeat CT timing, diagnostic yield, ionizing radiation exposure, and predictors of repetitive imaging. Results Although 637 manuscripts and abstracts were identified, no direct evidence was found. Thirteen documents were synthesized as indirect evidence. None of the indirect evidence defined a low-risk subset of abdominal pain, nor did investigators describe whether re-imaging occurred for complaints similar to the initial ED evaluation. Included studies did not describe the index CT findings and some reported explanatory findings noted on the original CT for which repeat CTs might have been indicated. The timeframe for a repeat CT ranged from hours to one year. The frequency of repeat CTs (2-47%) varied across studies as did the yield of imaging to alter downstream clinical decision-making (range 5-67%). Conclusion Due to the absence of direct evidence our scoping review is unable to provide high-quality evidence-based recommendations upon which to confidently base an imaging practice guideline. There is no evidence to support or refute performing a CT for low-risk recurrent abdominal pain.