Rajkumar Dhar
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
Rajkumar Dhar has a total of 39 co-authors. Here are the top co-authors by number of shared publications:
Hansen A Mansy
University of Central Florida, 4000 Central Florida Blvd, Orlando, 32816, UNITED STATES
5 shared publicationsRichard H Sandler
University of Central Florida, 4000 Central Florida Blvd, Orlando, Florida, 32816-2368, UNITED STATES
5 shared publicationsKim Manwaring
Pediatric Neurosurgery, Arnold Palmer Children’s Hospital, Orlando, USA
3 shared publicationsPeshala T Gamage
Florida Institute of Technology, 150 W University Blvd, Melbourne, Florida, 32901-8995, UNITED STATES
3 shared publications- Hansen A. Mansy
Biomedical Acoustics Research Laboratory, University of Central Florida, 4000 Central Florida Blvd, Orlando, FL 32816, USA
2 shared publications Nathan Kostick
College of Medicine, University of Central Florida, Orlando, USA
2 shared publications
Latest Publications
- Open Access
- 10 November 2026
- Published by Houston Methodist DeBakey Heart & Vascular Center in Methodist DeBakey Cardiovascular Journal
- Vol. 22 (5) , 73-86
Artificial Intelligence in Congenital Heart Disease: A Roadmap
Congenital heart disease (CHD) is a lifelong condition that requires ongoing monitoring due to its heterogeneous anatomy, physiology, and clinical course. Despite major advances in diagnosis and treatment, challenges remain in early detection, risk stratification, and long-term management. Artificial intelligence (AI) has the potential to enhance CHD care by extracting actionable insights from complex clinical data. This review examines current and emerging applications of AI in three key domains: cardiovascular imaging, wearable technologies, and clinical registries. We highlight opportunities to improve diagnosis, monitoring, and outcome prediction, while discussing important challenges related to data quality, validation, interpretability, and clinical implementation. Together, these areas illustrate both the promise and the barriers to integrating AI into CHD care.
- 10 November 2026
- Published by Houston Methodist DeBakey Heart & Vascular Center in Methodist DeBakey Cardiovascular Journal
- Vol. 22 (5) , 73-86
Open AccessArtificial Intelligence in Congenital Heart Disease: A Roadmap
Congenital heart disease (CHD) is a lifelong condition that requires ongoing monitoring due to its heterogeneous anatomy, physiology, and clinical course. Despite major advances in diagnosis and treatment, challenges remain in early detection, risk stratification, and long-term management. Artificial intelligence (AI) has the potential to enhance CHD care by extracting actionable insights from complex clinical data. This review examines current and emerging applications of AI in three key domains: cardiovascular imaging, wearable technologies, and clinical registries. We highlight opportunities to improve diagnosis, monitoring, and outcome prediction, while discussing important challenges related to data quality, validation, interpretability, and clinical implementation. Together, these areas illustrate both the promise and the barriers to integrating AI into CHD care.
- Open Access
- 4 November 2025
- Published by IOP Publishing in Physiological Measurement
- Vol. 46 (11) , 115003
AI-based approach for heart failure readmission prediction using SCG, ECG, and GSR signals
Objective. Heart failure (HF) is considered a global pandemic because of increasing prevalence, high mortality rate, frequent hospitalization, and associated economic burden. This study explores a noninvasive method that may help in managing HF patients by predicting HF readmission. Methods. Seismocardiogram (SCG) signal is the low-frequency chest vibration produced by the mechanical activity of the heart. SCG signal was acquired from 101 patients with HF, including those readmitted to the hospital during the study period. SCG signals were segmented into heartbeats and clustered based on respiration phases. Features were extracted from each cluster. Several conventional machine learning (ML) models were developed using selected SCG and heart rate variability features. Furthermore, SCG signals were transformed into images using a time–frequency distribution method. Images were used to train a deep learning model. The models were able to predict the readmission status of HF patients. Results. ML algorithms achieved higher accuracy than the deep learning model in classifying the readmitted and non-readmitted HF patients. K-nearest neighbor achieved the highest classification accuracy (89.4% accuracy, 87.8% sensitivity, 90.1% specificity, 78.2% precision, and 82.7% F 1-score). A detailed discussion of the extracted features was provided, correlating them with HF conditions. Conclusions . The study results suggest that SCG signals may be useful for readmission prediction of HF patients.
- 4 November 2025
- Published by IOP Publishing in Physiological Measurement
- Vol. 46 (11) , 115003
Open AccessAI-based approach for heart failure readmission prediction using SCG, ECG, and GSR signals
Objective. Heart failure (HF) is considered a global pandemic because of increasing prevalence, high mortality rate, frequent hospitalization, and associated economic burden. This study explores a noninvasive method that may help in managing HF patients by predicting HF readmission. Methods. Seismocardiogram (SCG) signal is the low-frequency chest vibration produced by the mechanical activity of the heart. SCG signal was acquired from 101 patients with HF, including those readmitted to the hospital during the study period. SCG signals were segmented into heartbeats and clustered based on respiration phases. Features were extracted from each cluster. Several conventional machine learning (ML) models were developed using selected SCG and heart rate variability features. Furthermore, SCG signals were transformed into images using a time–frequency distribution method. Images were used to train a deep learning model. The models were able to predict the readmission status of HF patients. Results. ML algorithms achieved higher accuracy than the deep learning model in classifying the readmitted and non-readmitted HF patients. K-nearest neighbor achieved the highest classification accuracy (89.4% accuracy, 87.8% sensitivity, 90.1% specificity, 78.2% precision, and 82.7% F 1-score). A detailed discussion of the extracted features was provided, correlating them with HF conditions. Conclusions . The study results suggest that SCG signals may be useful for readmission prediction of HF patients.
- 21 July 2025
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 196 (Pt B) , 110787
Automated detection of non-physiological artifacts on ECG signal: UK Biobank and CRIC
- 21 July 2025
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 196 (Pt B) , 110787
Automated detection of non-physiological artifacts on ECG signal: UK Biobank and CRIC
- 21 December 2024
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 185, 109600
Effect of respiration and exercise on seismocardiographic signals
- 21 December 2024
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 185, 109600
Effect of respiration and exercise on seismocardiographic signals
- Open Access
- 6 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Heart Failure Readmission Prediction Using Seismocardiogram Signal
Heart failure (HF) is considered a global pandemic because of increasing prevalence, high mortality rate, frequent hospitalization and associated economic burden. This study explores a noninvasive method that may help in managing HF patients by predicting HF readmission. Methods: Seismocardiogram (SCG) signal is the low frequency chest vibration produced by the mechanical activity of heart. SCG signal was acquired from 101 patients with HF including in those readmitted to the hospital during the study period. Features were extracted from SCG signals. Several conventional machine learning (ML) models were developed using selected SCG and heart rate variability features. Furthermore, SCG signals were transformed into images using a time-frequency distribution method. Images were used to train a deep learning model. The models were able to predict the readmission status of the HF patients. Results: ML algorithms achieved higher accuracy than the deep learning model in classifying the readmitted and non-readmitted HF patients. K-nearest neighbor (KNN) achieved the highest classification accuracy (89.4% accuracy, 87.8% sensitivity, and 90.1% specificity). The study results suggest that SCG signal may be useful for readmission prediction of HF patients. Significance: Use of SCG signal may help the management of HF patients and improve their quality of life.
- 6 March 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessHeart Failure Readmission Prediction Using Seismocardiogram Signal
Heart failure (HF) is considered a global pandemic because of increasing prevalence, high mortality rate, frequent hospitalization and associated economic burden. This study explores a noninvasive method that may help in managing HF patients by predicting HF readmission. Methods: Seismocardiogram (SCG) signal is the low frequency chest vibration produced by the mechanical activity of heart. SCG signal was acquired from 101 patients with HF including in those readmitted to the hospital during the study period. Features were extracted from SCG signals. Several conventional machine learning (ML) models were developed using selected SCG and heart rate variability features. Furthermore, SCG signals were transformed into images using a time-frequency distribution method. Images were used to train a deep learning model. The models were able to predict the readmission status of the HF patients. Results: ML algorithms achieved higher accuracy than the deep learning model in classifying the readmitted and non-readmitted HF patients. K-nearest neighbor (KNN) achieved the highest classification accuracy (89.4% accuracy, 87.8% sensitivity, and 90.1% specificity). The study results suggest that SCG signal may be useful for readmission prediction of HF patients. Significance: Use of SCG signal may help the management of HF patients and improve their quality of life.
- 10 February 2023
- Published by Springer Nature
- p. 121-147
Non-invasive ICP Monitoring by Auditory System Measurements
- 10 February 2023
- Published by Springer Nature
- p. 121-147
Non-invasive ICP Monitoring by Auditory System Measurements
- Open Access
- 1 February 2023
- Published by IOP Publishing in Physiological Measurement
- Vol. 44 (2) , 025001
Postural and longitudinal variability in seismocardiographic signals
Objective . Low frequency cardiovascular vibrations detectable on the chest surface (termed seismocardiography or SCG) may be useful for non-invasive diagnosis and monitoring of various cardiovascular conditions. A potential limitation of using SCG for longitudinal patient monitoring is the existence of intra-subject variability, which can contribute to errors in calculating SCG features. Improved understanding of the contribution of intra-subject variability sources may lead to improved SCG utility. This study aims to quantify postural and longitudinal SCG variability in healthy resting subjects during normal breathing. Approach . SCG and ECG signals were longitudinally acquired in 19 healthy subjects at different postures (supine, 45° head up, and sitting) during five recording sessions over five months. SCG cycles were segmented using the ECG R wave. Unsupervised machine learning was used to reduce SCG variability due to respiration by grouping the SCG signals into two clusters with minimized intra-cluster waveform heterogeneity. Several SCG features were assessed at different postures and longitudinally. Main results . SCG waveform morphological variability was calculated within each cluster (intra-cluster) and between two clusters (inter-cluster) at each posture and data collection session. The variabilities were significantly different between the supine and sitting but not between supine and 45° postures. For the 45° and sitting postures, the intra-cluster variability was not significantly different, while the inter-cluster variability difference was significant. The energy ratio between different frequency bands to total spectral energy in 0.5–50 Hz were calculated and were comparable for all postures. The combined cardiac timing intervals from the two clusters showed significant variation with postural changes. There was significant heart rate difference between the clusters and between postural positions. The SCG features were compared between longitudinal sessions and all features were not significantly different, Significance . Several SCG features significantly varied with posture suggesting that posture needs to be specified when comparing SCG changes over time. Longitudinally comparable SCG feature values suggests that significant longitudinal differences, if observed, may reflect true alternations in the cardiac functioning over time.
- 1 February 2023
- Published by IOP Publishing in Physiological Measurement
- Vol. 44 (2) , 025001
Open AccessPostural and longitudinal variability in seismocardiographic signals
Objective . Low frequency cardiovascular vibrations detectable on the chest surface (termed seismocardiography or SCG) may be useful for non-invasive diagnosis and monitoring of various cardiovascular conditions. A potential limitation of using SCG for longitudinal patient monitoring is the existence of intra-subject variability, which can contribute to errors in calculating SCG features. Improved understanding of the contribution of intra-subject variability sources may lead to improved SCG utility. This study aims to quantify postural and longitudinal SCG variability in healthy resting subjects during normal breathing. Approach . SCG and ECG signals were longitudinally acquired in 19 healthy subjects at different postures (supine, 45° head up, and sitting) during five recording sessions over five months. SCG cycles were segmented using the ECG R wave. Unsupervised machine learning was used to reduce SCG variability due to respiration by grouping the SCG signals into two clusters with minimized intra-cluster waveform heterogeneity. Several SCG features were assessed at different postures and longitudinally. Main results . SCG waveform morphological variability was calculated within each cluster (intra-cluster) and between two clusters (inter-cluster) at each posture and data collection session. The variabilities were significantly different between the supine and sitting but not between supine and 45° postures. For the 45° and sitting postures, the intra-cluster variability was not significantly different, while the inter-cluster variability difference was significant. The energy ratio between different frequency bands to total spectral energy in 0.5–50 Hz were calculated and were comparable for all postures. The combined cardiac timing intervals from the two clusters showed significant variation with postural changes. There was significant heart rate difference between the clusters and between postural positions. The SCG features were compared between longitudinal sessions and all features were not significantly different, Significance . Several SCG features significantly varied with posture suggesting that posture needs to be specified when comparing SCG changes over time. Longitudinally comparable SCG feature values suggests that significant longitudinal differences, if observed, may reflect true alternations in the cardiac functioning over time.
- Open Access
- 9 November 2021
- Published by Springer Nature in Scientific Reports
- Vol. 11 (1) , 1-13
Noninvasive detection of elevated ICP using spontaneous tympanic membrane pulsation
Neurological conditions such as traumatic brain injury (TBI) and hydrocephalus may lead to intracranial pressure (ICP) elevation. Current diagnosis methods rely on direct pressure measurement, while CT, MRI and other expensive imaging may be used. However, these invasive or expensive testing methods are often delayed because symptoms of elevated ICP are non-specific. Invasive methods, such as intraventricular catheter, subdural screw, epidural sensor, lumbar puncture, are associated with an increased risk of infection and hemorrhage. On the other hand, noninvasive, low-cost, accurate methods of ICP monitoring can help avoid risks and reduce costs while expediting diagnosis and treatment. The current study proposes and evaluates a novel method for noninvasive ICP monitoring using tympanic membrane pulsation (TMp). These signals are believed to be transmitted from ICP to the auditory system through the cochlear aqueduct. Fifteen healthy subjects were recruited and TMp signals were acquired noninvasively while the subjects performed maneuvers that are known to change ICP. A custom made system utilizing a stethoscope headset and a pressure transducer was used to perform these measurements. Maneuvers included head-up-tilt, head-down-tilt and hyperventilation. When elevated ICP was induced, significant TMp waveform morphological changes were observed in each subject (p < 0.01). These changes include certain waveform slopes and high frequency wave features. The observed changes were reversed by the maneuvers that decreased ICP (p < .01). The study results suggest that TMp waveform measurement and analysis may offer an inexpensive, noninvasive, accurate tool for detection and monitoring of ICP elevations. Further studies are warranted to validate this technique in patients with pathologically elevated ICP.
- 9 November 2021
- Published by Springer Nature in Scientific Reports
- Vol. 11 (1) , 1-13
Open AccessNoninvasive detection of elevated ICP using spontaneous tympanic membrane pulsation
Neurological conditions such as traumatic brain injury (TBI) and hydrocephalus may lead to intracranial pressure (ICP) elevation. Current diagnosis methods rely on direct pressure measurement, while CT, MRI and other expensive imaging may be used. However, these invasive or expensive testing methods are often delayed because symptoms of elevated ICP are non-specific. Invasive methods, such as intraventricular catheter, subdural screw, epidural sensor, lumbar puncture, are associated with an increased risk of infection and hemorrhage. On the other hand, noninvasive, low-cost, accurate methods of ICP monitoring can help avoid risks and reduce costs while expediting diagnosis and treatment. The current study proposes and evaluates a novel method for noninvasive ICP monitoring using tympanic membrane pulsation (TMp). These signals are believed to be transmitted from ICP to the auditory system through the cochlear aqueduct. Fifteen healthy subjects were recruited and TMp signals were acquired noninvasively while the subjects performed maneuvers that are known to change ICP. A custom made system utilizing a stethoscope headset and a pressure transducer was used to perform these measurements. Maneuvers included head-up-tilt, head-down-tilt and hyperventilation. When elevated ICP was induced, significant TMp waveform morphological changes were observed in each subject (p < 0.01). These changes include certain waveform slopes and high frequency wave features. The observed changes were reversed by the maneuvers that decreased ICP (p < .01). The study results suggest that TMp waveform measurement and analysis may offer an inexpensive, noninvasive, accurate tool for detection and monitoring of ICP elevations. Further studies are warranted to validate this technique in patients with pathologically elevated ICP.
- Open Access
- 23 August 2021
- Published by Springer Nature in Research Square
Noninvasive Detection of Elevated ICP Using Spontaneous Tympanic Membrane Pulsation
Neurological conditions such as traumatic brain injury (TBI) and hydrocephalus may lead to intracranial pressure (ICP) elevation, which can result in severe headaches, blurred vision, moving or talking problems, seizures and even death. Due to these potential risks, prompt medical attention and reliable monitoring would be needed when elevated ICP is suspected. Current diagnosis methods rely on direct pressure measurement, while CT, MRI and other expensive imaging may be used. However, these invasive or expensive testing methods are often delayed because the above symptoms are non-specific. In addition, invasive methods, such as intraventricular catheter, subdural screw, epidural sensor, lumbar puncture, are associated with an increased risk of infection and hemorrhage. On the other hand, noninvasive, low-cost, accurate methods of ICP monitoring can help avoid risks and reduce costs while expediting diagnosis and treatment. The current study proposes and evaluates a novel method for noninvasive ICP monitoring using tympanic membrane pulsation (TMp). These signals are believed to be transmitted from ICP to the auditory system through the cochlear aqueduct. Fifteen healthy subjects were recruited and TMp signals were acquired noninvasively while the subjects performed maneuvers that are known to change ICP. When elevated ICP was induced, significant TMp waveform morphological changes were observed in each subject (p < 0.01). These changes include certain waveform slopes and high frequency wave features. The observed changes were reversed by the maneuvers that decreased ICP (p <.01). The study results suggest that TMp waveform measurement and analysis may offer an inexpensive, noninvasive, accurate tool for detection and monitoring of ICP elevations. Further studies are warranted to validate this technique in patients with pathologically elevated ICP.
- 23 August 2021
- Published by Springer Nature in Research Square
Open AccessNoninvasive Detection of Elevated ICP Using Spontaneous Tympanic Membrane Pulsation
Neurological conditions such as traumatic brain injury (TBI) and hydrocephalus may lead to intracranial pressure (ICP) elevation, which can result in severe headaches, blurred vision, moving or talking problems, seizures and even death. Due to these potential risks, prompt medical attention and reliable monitoring would be needed when elevated ICP is suspected. Current diagnosis methods rely on direct pressure measurement, while CT, MRI and other expensive imaging may be used. However, these invasive or expensive testing methods are often delayed because the above symptoms are non-specific. In addition, invasive methods, such as intraventricular catheter, subdural screw, epidural sensor, lumbar puncture, are associated with an increased risk of infection and hemorrhage. On the other hand, noninvasive, low-cost, accurate methods of ICP monitoring can help avoid risks and reduce costs while expediting diagnosis and treatment. The current study proposes and evaluates a novel method for noninvasive ICP monitoring using tympanic membrane pulsation (TMp). These signals are believed to be transmitted from ICP to the auditory system through the cochlear aqueduct. Fifteen healthy subjects were recruited and TMp signals were acquired noninvasively while the subjects performed maneuvers that are known to change ICP. When elevated ICP was induced, significant TMp waveform morphological changes were observed in each subject (p < 0.01). These changes include certain waveform slopes and high frequency wave features. The observed changes were reversed by the maneuvers that decreased ICP (p <.01). The study results suggest that TMp waveform measurement and analysis may offer an inexpensive, noninvasive, accurate tool for detection and monitoring of ICP elevations. Further studies are warranted to validate this technique in patients with pathologically elevated ICP.
- Open Access
- 13 March 2021
- Published by Springer Nature in Cureus
- Vol. 13 (3)
The "Brain Stethoscope": A Non-Invasive Method for Detecting Elevated Intracranial Pressure
Introduction Minimally invasive intracranial pressure (ICP) screening has long been desired by neurosurgeons. A novel approach deriving ICP from tympanic membrane (TM) pulsation may offer the solution. The ICP waveform appears to be transmitted to the TM by the cochlear aqueduct. The resulting TM infrasonic pulsations can be measured by certain sensors. Elevated ICP alters brain compliance, which appears to yield slower rise times of the TM pulsation waveform. Measurement of this change may be useful in screening for elevated ICP. This paper investigates one such technique. Methods A stethoscope was modified for airtight external ear canal fit; the dome was exchanged for a magnetic reluctance pressure sensor, allowing measurement of TM pulsations. Analog TM pulsations were analyzed by measuring the pulsation's slope ratio between the waveform's downslope and upslope. Seventeen normal subjects (ages 18-32 years) underwent hyperventilation and tilt table testing to induce ICP changes. An algorithm processed this data and predicted the subject's ICP status. Results The slope ratio method showed consistent and stable changes with the expected alterations in ICP from the tilt test and hyperventilation maneuvers. The classification algorithm correctly identified subjects with elevated ICP in 60 of 60 independent recordings on 17 subjects. Conclusion This paper has four conclusions. First, the "brain stethoscope" can detect increased ICP from the TM pulsation waveform in healthy subjects. Second, analysis of the TM waveform using slope ratio calculations is capable of distinguishing normal versus elevated ICP. Third, the tilt and hyperventilation maneuvers showed the expected physiologic trends. Last, further studies are needed on patients with pathological ICP before the brain stethoscope can be implemented into clinical practice.
- 13 March 2021
- Published by Springer Nature in Cureus
- Vol. 13 (3)
Open AccessThe "Brain Stethoscope": A Non-Invasive Method for Detecting Elevated Intracranial Pressure
Introduction Minimally invasive intracranial pressure (ICP) screening has long been desired by neurosurgeons. A novel approach deriving ICP from tympanic membrane (TM) pulsation may offer the solution. The ICP waveform appears to be transmitted to the TM by the cochlear aqueduct. The resulting TM infrasonic pulsations can be measured by certain sensors. Elevated ICP alters brain compliance, which appears to yield slower rise times of the TM pulsation waveform. Measurement of this change may be useful in screening for elevated ICP. This paper investigates one such technique. Methods A stethoscope was modified for airtight external ear canal fit; the dome was exchanged for a magnetic reluctance pressure sensor, allowing measurement of TM pulsations. Analog TM pulsations were analyzed by measuring the pulsation's slope ratio between the waveform's downslope and upslope. Seventeen normal subjects (ages 18-32 years) underwent hyperventilation and tilt table testing to induce ICP changes. An algorithm processed this data and predicted the subject's ICP status. Results The slope ratio method showed consistent and stable changes with the expected alterations in ICP from the tilt test and hyperventilation maneuvers. The classification algorithm correctly identified subjects with elevated ICP in 60 of 60 independent recordings on 17 subjects. Conclusion This paper has four conclusions. First, the "brain stethoscope" can detect increased ICP from the TM pulsation waveform in healthy subjects. Second, analysis of the TM waveform using slope ratio calculations is capable of distinguishing normal versus elevated ICP. Third, the tilt and hyperventilation maneuvers showed the expected physiologic trends. Last, further studies are needed on patients with pathological ICP before the brain stethoscope can be implemented into clinical practice.