A control engineer with over 40 years of experience in advanced control, Guy Dumont has for the last 18 years focused on applications of signal processing and control to biomedical applications with a focus on critical care and mobile health.
The design of paper machine cross directional (CD) control systems is normally based on actuator ... more The design of paper machine cross directional (CD) control systems is normally based on actuator response models obtained in the centre rather than the machine edges. However process characteristics near the sheet edges are different from those in the centre of the sheet. Even with good CD control, the edge profile often shows the greatest variations from the target level. The application of control algorithms more suited to the centre of the machine reduces controller effectiveness at sheet edges. Control close to the sheet edge is normally carried out using a variety of assumptions concerning the expected response, however, these assumptions are often made for mathematical convenience rather than on the basis of physical principles. This work involves the comparison of edge response of the cross directional slice lip actuators with the response at the centre of the paper machine. The performance of a typical industrial CD actuator centre response model is tested and found to be inadequate when modeling the edge response. A new physical model, using surface wave theory of the slurry is investigated, tested and later modified in an attempt to better model the edge response. The new edge model is found to generate an improved edge response prediction, based on the centre response, but in general will be computationally expensive.
Modeling is fundamental to both feed-forward and feedback control. Within automated anesthesia th... more Modeling is fundamental to both feed-forward and feedback control. Within automated anesthesia the two paradigms are usually referred to as targetcontrolled infusion (TCI) and closed-loop drug delivery, respectively. In both cases, the objective is to control a system with anesthetic drug infusion rate as input, and (measured) clinical effect as output. The input is related to the output through the pharmacokinetics (PK) and pharmacodynamics (PD) of the patient. This chapter gives an introduction to PKPD modeling in automated anesthesia management, intended to be accessible to both anesthesiology and (control) engineering researchers. The following topics are discussed: the role of modeling; the classic PKPD structure used in clinical pharmacology; anesthesia modeling and identification for closedloop control; inter-patient variability and model uncertainty; disturbance, noise and equipment models. The chapter emphasizes electroencephalogram-guided control of propofol.
The COVID-19 pandemic has had an enormous toll on human health and well-being and led to major so... more The COVID-19 pandemic has had an enormous toll on human health and well-being and led to major social and economic disruptions. Public health interventions in response to burgeoning case numbers and hospitalizations have repeatedly bent down the epidemic curve, effectively creating a feedback control system. Worst case scenarios have been avoided in many places through this responsive feedback. We aim to formalize and illustrate how to incorporate principles of feedback control into pandemic projections and decision-making, and ultimately shift the focus from prediction to the design of interventions. Starting with an epidemiological model for COVID-19, we illustrate how feedback control can be incorporated into pandemic management using a simple design that couples recent changes in case numbers or hospital occupancy with explicit policy restrictions. We demonstrate robust ability to control a pandemic using a design that responds to hospital cases, despite simulating large uncertainty in reproduction number R 0 (range: 1.04-5.18) and average time to hospital admission (range: 4-28 days). We show that shorter delays, responding to case counts versus hospital measured infections, reduce both the cumulative case count and the average level of interventions. Finally, we show that feedback is robust to changing compliance to public health directives and to systemic changes associated with variants of concern and with the introduction of a vaccination program. The negative impact of a pandemic on human health and societal disruption can be reduced by coupling models of disease propagation with models of the decision-making process. In contrast to highly varying open-loop projections, incorporating feedback explicitly in the decision-making process is more reflective of the real-world challenge facing public health decision makers. Using feedback principles, effective control strategies can be designed even if the pandemic characteristics are highly uncertain, encouraging earlier and smaller actions that reduce both case counts and the extent of interventions.
CLOSED FOR BUSINESS: The streets of Manhattan are quiet on 10 April as most people comply with th... more CLOSED FOR BUSINESS: The streets of Manhattan are quiet on 10 April as most people comply with the city's self-quarantine rules.
This master thesis project proposes methods for individualizing closed-loop controlled anesthesia... more This master thesis project proposes methods for individualizing closed-loop controlled anesthesia. One of the largest challenges with closed-loop anesthesia is the variation between patients in the sensitivity to the anesthetic drug, here propofol. Due to limited excitation in the process dynamics together with a high measurement noise level is it not possible to determine a full reliable model describing a patient's dynamics online. The method used here for minimizing the effects of inter-patient variability was through patient model partitioning of children and adult models. Partitioning was based on similarity measures between patients, for example age, weight and applied to a dynamic model describing each patient. For each subset resulting from partitioning, an optimal PID controller has been synthesized. This thesis has shown that the effects of inter-patient variability can be reduced using partitioning into two subsets. More subsets did not result in a significant reduction. Partitioning based on ν-gap between patient models resulted in the best attenuation of surgical stimulation disturbances. Partitioning based on age for children and weight for adults reduces the impact from surgical stimulation were proposed for clinical practices. These methods are easy to implement because the demographics are known beforehand and does not depend on actual measurements during the anesthesia. The results are substantiated by simulations and calculations of achieved attenuation with acceptable performance and preserved robustness. I would first like to thank my supervisor Kristian Soltesz. I am very grateful to you for introducing me to this interesting research field. You have always come with valuable inputs on my proceeding work and I am very grateful for all of your help. In the research group at BCCHR, University of British Columbia, Vancouver, Canada, I must acknowledge professor Guy Dumont and Dr. Klaske van Heusden for letting me visit and take part of their work at the Childrens Hospital of British Columbia. I would also like to thank Dr. Mark Ansermino for letting me visit anesthesia surgeries during my stay, giving me a broader perspective of the subject. The rest of the research group requires an acknowledgement for the warm welcome during my stay in Vancouver. I also want to thank researcher Richard Pates for the introduction and help with implementation of the ν-gap. A special thank to PhD student José Manuel González Cava who has provided me with the code to the synthesis, particularly designed for the children model set. He has also been very helpful during the project, always available for questions. 5
Biophotonics in Exercise Science, Sports Medicine, Health Monitoring Technologies, and Wearables
Electroencephalography (EEG) and cerebral near-infrared spectroscopy (NIRS) are both well-known m... more Electroencephalography (EEG) and cerebral near-infrared spectroscopy (NIRS) are both well-known monitoring methods to quantify cerebral neurophysiology and hemodynamics states of the brain. A stable regulatory system operates to guarantee sufficient spatial and temporal distribution of energy substrates for ongoing neuronal activity. Most EEG signals are associated with the neural activity of an enormous number of neurons that are interconnected and firing concurrently. The conventional EEG bandwidth is 0.16Hz to 70Hz. In this study, the EEG recording bandwidth is extended in low frequency (0.016Hz to 70Hz) by using a novel EEG amplifier. We aimed to investigate the low-frequency EEG and brain tissue deoxygenation by using novel multi-modal measurements. We used combined NIRS and EEG measurements for estimating the electrophysiological activity and hemodynamic changes in the adult human forehead during a hypoxic breathing condition. For the experiment, an altitude simulation kit was used to restrict the concentration of oxygen in the air that was inhaled by the subjects. The hypoxic breathing conditions led to variations in CO2 concentration (pCO2). Prolong (low-frequency) EEG signal shift, accompanied by an increase of deoxygenated hemoglobin during simulated hypoxic breathing were observed in this experiment.
This paper explores automated face and facial landmark detection of neonates, which is an importa... more This paper explores automated face and facial landmark detection of neonates, which is an important first step in many video-based neonatal health applications, such as vital sign estimation, pain assessment, sleep-wake classification, and jaundice detection. Utilising three publicly available datasets of neonates in the clinical environment, 366 images (258 subjects) and 89 (66 subjects) were annotated for training and testing, respectively. Transfer learning was applied to two YOLO-based models, with input training images augmented with random horizontal flipping, photo-metric colour distortion, translation and scaling during each training epoch. Additionally, the re-orientation of input images and fusion of trained deep learning models was explored. Our proposed model based on YOLOv7Face outperformed existing methods with a mean average precision of 84.8% for face detection, and a normalised mean error of 0.072 for facial landmark detection. Overall, this will assist in the development of fully automated neonatal health assessment algorithms. Clinical relevance-Accurate face and facial landmark detection provides an automated and non-contact option to assist in video-based neonatal health applications.
Optics and Biophotonics in Low-Resource Settings V, 2019
According to the WHO, 15,000 children under five years are dying every day from preventable cause... more According to the WHO, 15,000 children under five years are dying every day from preventable causes with 80% of these children being born in low-income countries. Portable optical medical diagnostic devices can help physicians, nurses and untrained health workers to objectively identify children who are at a higher risk of dying. In the last 2 years, we collected the oxygenation values of the brachioradialis muscle, using a commercial Near Infrared Spectroscopy (NIRS) device, in 200 children under 5 years admitted in two hospitals in Uganda. Data revealed that the tissue oxygen saturation decrease during a vascular occlusion predicts children at higher risk better than other vital signs (SpO2, respiration rate, heart rate and temperature). Based on these results, we designed a low cost Continuous Wave Spatially Resolved NIRS device controlled by a smartphone in order to extend our study to a larger population and confirm our observation. The total cost of this device (excluding the smartphone) is less than $100. The preliminary tests suggest a significant potential of our low cost mobile NIRS device and oxygenation values closely matching those reported by the best device on the market.
We propose to use noncausal transfer functions to model the spatial behavior of cross-directional... more We propose to use noncausal transfer functions to model the spatial behavior of cross-directional (CD) processes so as to circumvent the high-dimensionality of a causal transfer function. This noncausal representation is shown to have a causal-equivalent form. We prove that the covariance of maximum likelihood estimate of the causal-equivalent model asymptotically converges to that of the noncausal model. This result is then used to design optimal inputs in closed-loop for the original noncausal model of the CD process. An illustrative example is provided to highlight the advantage of using optimally designed excitation signal for CD closed-loop identification over white noise excitation or the current industrial practice of spatial bump excitation.
Introduction Respiration can modulate the intensity, frequency and amplitude of the photoplethysm... more Introduction Respiration can modulate the intensity, frequency and amplitude of the photoplethysmogram (PPG) waveform [1]. During respiration, the pressure variation in the thoracic cavity causes blood to be exchanged with the pulmonary circuit, resulting in a baseline perfusion variation (intensity variation). There is a corresponding decrease in cardiac output due to reduced ventricular filling, causing an amplitude variation in the peripheral pulse. Also, there is an autonomic response, where the heart rate (HR) variation synchronizes with the respiratory cycle, causing a respiratory induced frequency variation of the PPG waveform. Physical experiments demonstrate varying levels of correlation between the individual respiratory-induced variations of the PPG signal and the reference respiratory measurements [1]. To solve the STA Annual Meeting 2012 Engineering Challenge, we suggest improving the predictive value of the non-invasive estimation of respiratory rate (RR) from the PPG ...
Background Blood pressure measurement is a marker of antenatal care quality. In well resourced se... more Background Blood pressure measurement is a marker of antenatal care quality. In well resourced settings, lower blood pressure cutoffs for hypertension are associated with adverse pregnancy outcomes. We aimed to study the associations between blood pressure thresholds and adverse outcomes and the diagnostic test properties of these blood pressure cutoffs in low-resource settings. We did a secondary analysis of data from 22 intervention clusters in the Community-Level Interventions for Pre-eclampsia (CLIP) cluster randomised trials (NCT01911494) in India (n=6), Mozambique (n=6), and Pakistan (n=10). We included pregnant women aged 15-49 years (12-49 years in Mozambique), identified in their community by trained community health workers, who had data on blood pressure measurements and outcomes. The trial was unmasked. Maximum blood pressure was categorised as: normal blood pressure (systolic blood pressure [sBP] <120 mm Hg and diastolic blood pressure [dBP] <80 mm Hg), elevated blood pressure (sBP 120-129 mm Hg and dBP <80 mm Hg), stage 1 hypertension (sBP 130-139 mm Hg or dBP 80-89 mm Hg, or both), non-severe stage 2 hypertension (sBP 140-159 mm Hg or dBP 90-109 mm Hg, or both), or severe stage 2 hypertension (sBP ≥160 mm Hg or dBP ≥110 mm Hg, or both). We classified women according to the maximum blood pressure category reached across all visits for the primary analyses. The primary outcome was a maternal, fetal, or neonatal mortality or morbidity composite. We estimated dose-response relationships between blood pressure category and adverse outcomes, as well as diagnostic test properties.
Background Incomplete vital registration systems mean that causes of death during pregnancy and c... more Background Incomplete vital registration systems mean that causes of death during pregnancy and childbirth are poorly understood in low-income and middle-income countries. To inform global efforts to reduce maternal mortality, we compared physician review and computerised analysis of verbal autopsies (interpreting verbal autopsies [InterVA] software), to understand their agreement on maternal cause of death and circumstances of mortality categories (COMCATs) in the Community-Level Interventions for Pre-eclampsia (CLIP) cluster randomised trials. Methods The CLIP trials took place in India, Pakistan, and Mozambique, enrolling pregnant women aged 12-49 years between Nov 1, 2014, and Feb 28, 2017. 69 330 pregnant women were enrolled in 44 clusters (36 008 in the 22 intervention clusters and 33 322 in the 22 control clusters). In this secondary analysis of maternal deaths in CLIP, we included women who died in any of the 22 intervention clusters or 22 control clusters. Trained staff administered the WHO 2012 verbal autopsy after maternal deaths. Two physicians (and a third for consensus, if needed) reviewed trial surveillance data and verbal autopsies, and, in intervention clusters, community health worker-led visit data. They determined cause of death according to the WHO International Classification of Diseases-Maternal Mortality (ICD-MM). Verbal autopsies were also analysed by InterVA computer models (versions 4 and 5) to generate cause of death. COMCAT analysis was provided by InterVA-5 and, in India, by physician review of Maternal Newborn Health Registry data. Causes of death and COMCATs assigned by physician review, Inter-VA-4, and InterVA-5 were compared, with agreement assessed with Cohen's κ coefficient. Findings Of 61 988 pregnancies with successful follow-up in the CLIP trials, 143 maternal deaths were reported (16 deaths in India, 105 in Pakistan, and 22 in Mozambique). The maternal death rate was 231 (95% CI 193-268) per 100 000 identified pregnancies. Most deaths were attributed to direct maternal causes (rather than indirect or undetermined causes as per ICD-MM classification), with fair to good agreement between physician review and InterVA-4 (κ=0•56 [95% CI 0•43-0•66]) or InterVA-5 (κ=0•44 [0•30-0•57]), and InterVA-4 and InterVA-5 (κ=0•72 [0•60-0•84]). The top three causes of death were the same by physician review, InterVA-4, and InterVA-5 (ICD-MM categories obstetric haemorrhage, non-obstetric complications, and hypertensive disorders); however, attribution of individual patient deaths to obstetric haemorrhage varied more between methods (physician review, 38 [27%] deaths; InterVA-4, 69 [48%] deaths; and InterVA-5, 82 [57%] deaths), than did attribution to non-obstetric causes (physician review, 39 [27%] deaths; InterVA-4, 37 [26%] deaths; and InterVA-5, 28 [20%] deaths) or hypertensive disorders (physician review, 23 [16%] deaths; InterVA-4, 25 [17%] deaths; and InterVA-5, 24 [17%] deaths). Agreement for all nine ICD-MM categories was fair for physician review versus InterVA-4 (κ=0•48 [0•38-0•58]), poor for physician review versus InterVA-5 (κ=0•36 [0•27-0•46]), and good for InterVA-4 versus InterVA-5 (κ=0•69 [0•59-0•79]). The most commonly assigned COMCATs by InterVA-5 were emergencies (68 [48%] of 143 deaths) and health systems (62 [43%] deaths), and by physician review (India only) were health systems (seven [44%] of 16 deaths) and inevitability (five [31%] deaths); agreement between InterVA-5 and physician review (India data only) was poor (κ=0•04 [0•00-0•15]). Interpretation Our findings indicate that InterVA-5 is less accurate than InterVA-4 at ascertaining causes and circumstances of maternal death, when compared with physician review. Our results suggest a need to improve the next iteration of InterVA, and for researchers and clinicians to preferentially use InterVA-4 when recording maternal deaths. Funding University of British Columbia (grantee of the Bill & Melinda Gates Foundation).
Telehealth is a strategy to expand the reach of pulmonary rehabilitation (PR). Smartphones can mo... more Telehealth is a strategy to expand the reach of pulmonary rehabilitation (PR). Smartphones can monitor and transmit oxygen saturation (SpO 2 ) and heart rate (HR) data to ensure patient safety during home-based or other exercise. The purpose of this study was to evaluate the usability, validity and reliability of a Kenek O 2 pulse oximeter and custom prototype smartphone application (smartphone oximeter) during rest and exercise in healthy participants and those with chronic lung disease. Methods Fifteen individuals with chronic lung disease and 15 healthy controls were recruited. SpO 2 and HR were evaluated at rest and during cycling and walking. SpO2 was valid if the mean bias was within +± 2%, the level of agreement (LoA) was within ± 4%; HR was valid if the mean bias was within ± 5 beats per min (bpm), LoA was within ± 10 bpm. Usability was assessed with a questionnaire and direct observation. The smartphone oximeter was deemed easy to use. At rest, SpO 2 measures were valid in both groups (bias <2%, lower bound LoA -2 to 3%). During exercise, SpO 2 measurement did not meet validity and reliability thresholds in the patients with chronic lung disease, but was accurate for the healthy controls. HR recording during exercise or rest was not valid (LoA > 10 bpm) in either group. Conclusions The smartphone oximeter did not record HR or SpO 2 accurately in patients with chronic lung disease during exercise, although SpO 2 was valid at rest. During exercise, patients with chronic lung disease should pause to ensure greatest accuracy of SpO 2 and HR measurement.
2022 44th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), Jul 11, 2022
Neonatal respiratory distress is a common condition that if left untreated, can lead to short-and... more Neonatal respiratory distress is a common condition that if left untreated, can lead to short-and long-term complications. This paper investigates the usage of digital stethoscope recorded chest sounds taken within 1 min post-delivery, to enable early detection and prediction of neonatal respiratory distress. Fifty-one term newborns were included in this study, 9 of whom developed respiratory distress. For each newborn, 1 min anterior and posterior recordings were taken. These recordings were pre-processed to remove noisy segments and obtain high-quality heart and lung sounds. The random undersampling boosting (RUSBoost) classifier was then trained on a variety of features, such as power and vital sign features extracted from the heart and lung sounds. The RUSBoost algorithm produced specificity, sensitivity, and accuracy results of 85.0%, 66.7% and 81.8%, respectively. Clinical relevance-This paper investigates the feasibility of digital stethoscope recorded chest sounds for early detection of respiratory distress in term newborn babies, to enable timely treatment and management. E.
In response to: Spurr R, Ng E, Onchiri FM, Rapha B, Nakatumba-Nabende J, Rosenfeld M, Najjingo I,... more In response to: Spurr R, Ng E, Onchiri FM, Rapha B, Nakatumba-Nabende J, Rosenfeld M, Najjingo I, Stout J, Nantanda R, Ellington LE. Performance and usability of a new mobile application for measuring respiratory rate in young children with acute lower respiratory infections. Pediatr Pulmonol. 2022 Aug 22. doi: 10.1002/ppul.26125.
IEEE Transactions on Biomedical Engineering, Oct 1, 2019
The goal of this study was to optimize robust PID control for propofol anesthesia in children age... more The goal of this study was to optimize robust PID control for propofol anesthesia in children aged 5-10 years to improve performance, particularly to decrease the time of induction of anesthesia while maintaining robustness. Methods: We analyzed results of a previous study conducted by our group to identify opportunities for system improvement. Allometric scaling was introduced to reduce the interpatient variability and a new robust PID controller was designed using an optimization based method. We evaluated this optimized design in a clinical study involving 16 new cases. Results: The optimized controller design achieved the performance predicted in simulation studies in the design stage. Time of induction of anesthesia was median [Q1, Q3] 3.7 [2.3, 4.1] minutes and the achieved global score was 13.4 [9.9, 16.8]. Conclusion: Allometric scaling reduces the interpatient variability in this age group, and allows for improved closed-loop performance. The uncertainty described by the model set, the predicted closedloop responses and the predicted robustness margins are realistic. The system meets the design objectives of improved speed of induction of anesthesia while maintaining robustness, improving clinically relevant system behavior. Significance: Control system optimization and ongoing system improvement are essential to the development of a clinically relevant commercial device. This paper demonstrates the validity of our approach, including system modeling, controller optimization and pre-clinical testing in simulation.
The design of paper machine cross directional (CD) control systems is normally based on actuator ... more The design of paper machine cross directional (CD) control systems is normally based on actuator response models obtained in the centre rather than the machine edges. However process characteristics near the sheet edges are different from those in the centre of the sheet. Even with good CD control, the edge profile often shows the greatest variations from the target level. The application of control algorithms more suited to the centre of the machine reduces controller effectiveness at sheet edges. Control close to the sheet edge is normally carried out using a variety of assumptions concerning the expected response, however, these assumptions are often made for mathematical convenience rather than on the basis of physical principles. This work involves the comparison of edge response of the cross directional slice lip actuators with the response at the centre of the paper machine. The performance of a typical industrial CD actuator centre response model is tested and found to be inadequate when modeling the edge response. A new physical model, using surface wave theory of the slurry is investigated, tested and later modified in an attempt to better model the edge response. The new edge model is found to generate an improved edge response prediction, based on the centre response, but in general will be computationally expensive.
Modeling is fundamental to both feed-forward and feedback control. Within automated anesthesia th... more Modeling is fundamental to both feed-forward and feedback control. Within automated anesthesia the two paradigms are usually referred to as targetcontrolled infusion (TCI) and closed-loop drug delivery, respectively. In both cases, the objective is to control a system with anesthetic drug infusion rate as input, and (measured) clinical effect as output. The input is related to the output through the pharmacokinetics (PK) and pharmacodynamics (PD) of the patient. This chapter gives an introduction to PKPD modeling in automated anesthesia management, intended to be accessible to both anesthesiology and (control) engineering researchers. The following topics are discussed: the role of modeling; the classic PKPD structure used in clinical pharmacology; anesthesia modeling and identification for closedloop control; inter-patient variability and model uncertainty; disturbance, noise and equipment models. The chapter emphasizes electroencephalogram-guided control of propofol.
The COVID-19 pandemic has had an enormous toll on human health and well-being and led to major so... more The COVID-19 pandemic has had an enormous toll on human health and well-being and led to major social and economic disruptions. Public health interventions in response to burgeoning case numbers and hospitalizations have repeatedly bent down the epidemic curve, effectively creating a feedback control system. Worst case scenarios have been avoided in many places through this responsive feedback. We aim to formalize and illustrate how to incorporate principles of feedback control into pandemic projections and decision-making, and ultimately shift the focus from prediction to the design of interventions. Starting with an epidemiological model for COVID-19, we illustrate how feedback control can be incorporated into pandemic management using a simple design that couples recent changes in case numbers or hospital occupancy with explicit policy restrictions. We demonstrate robust ability to control a pandemic using a design that responds to hospital cases, despite simulating large uncertainty in reproduction number R 0 (range: 1.04-5.18) and average time to hospital admission (range: 4-28 days). We show that shorter delays, responding to case counts versus hospital measured infections, reduce both the cumulative case count and the average level of interventions. Finally, we show that feedback is robust to changing compliance to public health directives and to systemic changes associated with variants of concern and with the introduction of a vaccination program. The negative impact of a pandemic on human health and societal disruption can be reduced by coupling models of disease propagation with models of the decision-making process. In contrast to highly varying open-loop projections, incorporating feedback explicitly in the decision-making process is more reflective of the real-world challenge facing public health decision makers. Using feedback principles, effective control strategies can be designed even if the pandemic characteristics are highly uncertain, encouraging earlier and smaller actions that reduce both case counts and the extent of interventions.
CLOSED FOR BUSINESS: The streets of Manhattan are quiet on 10 April as most people comply with th... more CLOSED FOR BUSINESS: The streets of Manhattan are quiet on 10 April as most people comply with the city's self-quarantine rules.
This master thesis project proposes methods for individualizing closed-loop controlled anesthesia... more This master thesis project proposes methods for individualizing closed-loop controlled anesthesia. One of the largest challenges with closed-loop anesthesia is the variation between patients in the sensitivity to the anesthetic drug, here propofol. Due to limited excitation in the process dynamics together with a high measurement noise level is it not possible to determine a full reliable model describing a patient's dynamics online. The method used here for minimizing the effects of inter-patient variability was through patient model partitioning of children and adult models. Partitioning was based on similarity measures between patients, for example age, weight and applied to a dynamic model describing each patient. For each subset resulting from partitioning, an optimal PID controller has been synthesized. This thesis has shown that the effects of inter-patient variability can be reduced using partitioning into two subsets. More subsets did not result in a significant reduction. Partitioning based on ν-gap between patient models resulted in the best attenuation of surgical stimulation disturbances. Partitioning based on age for children and weight for adults reduces the impact from surgical stimulation were proposed for clinical practices. These methods are easy to implement because the demographics are known beforehand and does not depend on actual measurements during the anesthesia. The results are substantiated by simulations and calculations of achieved attenuation with acceptable performance and preserved robustness. I would first like to thank my supervisor Kristian Soltesz. I am very grateful to you for introducing me to this interesting research field. You have always come with valuable inputs on my proceeding work and I am very grateful for all of your help. In the research group at BCCHR, University of British Columbia, Vancouver, Canada, I must acknowledge professor Guy Dumont and Dr. Klaske van Heusden for letting me visit and take part of their work at the Childrens Hospital of British Columbia. I would also like to thank Dr. Mark Ansermino for letting me visit anesthesia surgeries during my stay, giving me a broader perspective of the subject. The rest of the research group requires an acknowledgement for the warm welcome during my stay in Vancouver. I also want to thank researcher Richard Pates for the introduction and help with implementation of the ν-gap. A special thank to PhD student José Manuel González Cava who has provided me with the code to the synthesis, particularly designed for the children model set. He has also been very helpful during the project, always available for questions. 5
Biophotonics in Exercise Science, Sports Medicine, Health Monitoring Technologies, and Wearables
Electroencephalography (EEG) and cerebral near-infrared spectroscopy (NIRS) are both well-known m... more Electroencephalography (EEG) and cerebral near-infrared spectroscopy (NIRS) are both well-known monitoring methods to quantify cerebral neurophysiology and hemodynamics states of the brain. A stable regulatory system operates to guarantee sufficient spatial and temporal distribution of energy substrates for ongoing neuronal activity. Most EEG signals are associated with the neural activity of an enormous number of neurons that are interconnected and firing concurrently. The conventional EEG bandwidth is 0.16Hz to 70Hz. In this study, the EEG recording bandwidth is extended in low frequency (0.016Hz to 70Hz) by using a novel EEG amplifier. We aimed to investigate the low-frequency EEG and brain tissue deoxygenation by using novel multi-modal measurements. We used combined NIRS and EEG measurements for estimating the electrophysiological activity and hemodynamic changes in the adult human forehead during a hypoxic breathing condition. For the experiment, an altitude simulation kit was used to restrict the concentration of oxygen in the air that was inhaled by the subjects. The hypoxic breathing conditions led to variations in CO2 concentration (pCO2). Prolong (low-frequency) EEG signal shift, accompanied by an increase of deoxygenated hemoglobin during simulated hypoxic breathing were observed in this experiment.
This paper explores automated face and facial landmark detection of neonates, which is an importa... more This paper explores automated face and facial landmark detection of neonates, which is an important first step in many video-based neonatal health applications, such as vital sign estimation, pain assessment, sleep-wake classification, and jaundice detection. Utilising three publicly available datasets of neonates in the clinical environment, 366 images (258 subjects) and 89 (66 subjects) were annotated for training and testing, respectively. Transfer learning was applied to two YOLO-based models, with input training images augmented with random horizontal flipping, photo-metric colour distortion, translation and scaling during each training epoch. Additionally, the re-orientation of input images and fusion of trained deep learning models was explored. Our proposed model based on YOLOv7Face outperformed existing methods with a mean average precision of 84.8% for face detection, and a normalised mean error of 0.072 for facial landmark detection. Overall, this will assist in the development of fully automated neonatal health assessment algorithms. Clinical relevance-Accurate face and facial landmark detection provides an automated and non-contact option to assist in video-based neonatal health applications.
Optics and Biophotonics in Low-Resource Settings V, 2019
According to the WHO, 15,000 children under five years are dying every day from preventable cause... more According to the WHO, 15,000 children under five years are dying every day from preventable causes with 80% of these children being born in low-income countries. Portable optical medical diagnostic devices can help physicians, nurses and untrained health workers to objectively identify children who are at a higher risk of dying. In the last 2 years, we collected the oxygenation values of the brachioradialis muscle, using a commercial Near Infrared Spectroscopy (NIRS) device, in 200 children under 5 years admitted in two hospitals in Uganda. Data revealed that the tissue oxygen saturation decrease during a vascular occlusion predicts children at higher risk better than other vital signs (SpO2, respiration rate, heart rate and temperature). Based on these results, we designed a low cost Continuous Wave Spatially Resolved NIRS device controlled by a smartphone in order to extend our study to a larger population and confirm our observation. The total cost of this device (excluding the smartphone) is less than $100. The preliminary tests suggest a significant potential of our low cost mobile NIRS device and oxygenation values closely matching those reported by the best device on the market.
We propose to use noncausal transfer functions to model the spatial behavior of cross-directional... more We propose to use noncausal transfer functions to model the spatial behavior of cross-directional (CD) processes so as to circumvent the high-dimensionality of a causal transfer function. This noncausal representation is shown to have a causal-equivalent form. We prove that the covariance of maximum likelihood estimate of the causal-equivalent model asymptotically converges to that of the noncausal model. This result is then used to design optimal inputs in closed-loop for the original noncausal model of the CD process. An illustrative example is provided to highlight the advantage of using optimally designed excitation signal for CD closed-loop identification over white noise excitation or the current industrial practice of spatial bump excitation.
Introduction Respiration can modulate the intensity, frequency and amplitude of the photoplethysm... more Introduction Respiration can modulate the intensity, frequency and amplitude of the photoplethysmogram (PPG) waveform [1]. During respiration, the pressure variation in the thoracic cavity causes blood to be exchanged with the pulmonary circuit, resulting in a baseline perfusion variation (intensity variation). There is a corresponding decrease in cardiac output due to reduced ventricular filling, causing an amplitude variation in the peripheral pulse. Also, there is an autonomic response, where the heart rate (HR) variation synchronizes with the respiratory cycle, causing a respiratory induced frequency variation of the PPG waveform. Physical experiments demonstrate varying levels of correlation between the individual respiratory-induced variations of the PPG signal and the reference respiratory measurements [1]. To solve the STA Annual Meeting 2012 Engineering Challenge, we suggest improving the predictive value of the non-invasive estimation of respiratory rate (RR) from the PPG ...
Background Blood pressure measurement is a marker of antenatal care quality. In well resourced se... more Background Blood pressure measurement is a marker of antenatal care quality. In well resourced settings, lower blood pressure cutoffs for hypertension are associated with adverse pregnancy outcomes. We aimed to study the associations between blood pressure thresholds and adverse outcomes and the diagnostic test properties of these blood pressure cutoffs in low-resource settings. We did a secondary analysis of data from 22 intervention clusters in the Community-Level Interventions for Pre-eclampsia (CLIP) cluster randomised trials (NCT01911494) in India (n=6), Mozambique (n=6), and Pakistan (n=10). We included pregnant women aged 15-49 years (12-49 years in Mozambique), identified in their community by trained community health workers, who had data on blood pressure measurements and outcomes. The trial was unmasked. Maximum blood pressure was categorised as: normal blood pressure (systolic blood pressure [sBP] <120 mm Hg and diastolic blood pressure [dBP] <80 mm Hg), elevated blood pressure (sBP 120-129 mm Hg and dBP <80 mm Hg), stage 1 hypertension (sBP 130-139 mm Hg or dBP 80-89 mm Hg, or both), non-severe stage 2 hypertension (sBP 140-159 mm Hg or dBP 90-109 mm Hg, or both), or severe stage 2 hypertension (sBP ≥160 mm Hg or dBP ≥110 mm Hg, or both). We classified women according to the maximum blood pressure category reached across all visits for the primary analyses. The primary outcome was a maternal, fetal, or neonatal mortality or morbidity composite. We estimated dose-response relationships between blood pressure category and adverse outcomes, as well as diagnostic test properties.
Background Incomplete vital registration systems mean that causes of death during pregnancy and c... more Background Incomplete vital registration systems mean that causes of death during pregnancy and childbirth are poorly understood in low-income and middle-income countries. To inform global efforts to reduce maternal mortality, we compared physician review and computerised analysis of verbal autopsies (interpreting verbal autopsies [InterVA] software), to understand their agreement on maternal cause of death and circumstances of mortality categories (COMCATs) in the Community-Level Interventions for Pre-eclampsia (CLIP) cluster randomised trials. Methods The CLIP trials took place in India, Pakistan, and Mozambique, enrolling pregnant women aged 12-49 years between Nov 1, 2014, and Feb 28, 2017. 69 330 pregnant women were enrolled in 44 clusters (36 008 in the 22 intervention clusters and 33 322 in the 22 control clusters). In this secondary analysis of maternal deaths in CLIP, we included women who died in any of the 22 intervention clusters or 22 control clusters. Trained staff administered the WHO 2012 verbal autopsy after maternal deaths. Two physicians (and a third for consensus, if needed) reviewed trial surveillance data and verbal autopsies, and, in intervention clusters, community health worker-led visit data. They determined cause of death according to the WHO International Classification of Diseases-Maternal Mortality (ICD-MM). Verbal autopsies were also analysed by InterVA computer models (versions 4 and 5) to generate cause of death. COMCAT analysis was provided by InterVA-5 and, in India, by physician review of Maternal Newborn Health Registry data. Causes of death and COMCATs assigned by physician review, Inter-VA-4, and InterVA-5 were compared, with agreement assessed with Cohen's κ coefficient. Findings Of 61 988 pregnancies with successful follow-up in the CLIP trials, 143 maternal deaths were reported (16 deaths in India, 105 in Pakistan, and 22 in Mozambique). The maternal death rate was 231 (95% CI 193-268) per 100 000 identified pregnancies. Most deaths were attributed to direct maternal causes (rather than indirect or undetermined causes as per ICD-MM classification), with fair to good agreement between physician review and InterVA-4 (κ=0•56 [95% CI 0•43-0•66]) or InterVA-5 (κ=0•44 [0•30-0•57]), and InterVA-4 and InterVA-5 (κ=0•72 [0•60-0•84]). The top three causes of death were the same by physician review, InterVA-4, and InterVA-5 (ICD-MM categories obstetric haemorrhage, non-obstetric complications, and hypertensive disorders); however, attribution of individual patient deaths to obstetric haemorrhage varied more between methods (physician review, 38 [27%] deaths; InterVA-4, 69 [48%] deaths; and InterVA-5, 82 [57%] deaths), than did attribution to non-obstetric causes (physician review, 39 [27%] deaths; InterVA-4, 37 [26%] deaths; and InterVA-5, 28 [20%] deaths) or hypertensive disorders (physician review, 23 [16%] deaths; InterVA-4, 25 [17%] deaths; and InterVA-5, 24 [17%] deaths). Agreement for all nine ICD-MM categories was fair for physician review versus InterVA-4 (κ=0•48 [0•38-0•58]), poor for physician review versus InterVA-5 (κ=0•36 [0•27-0•46]), and good for InterVA-4 versus InterVA-5 (κ=0•69 [0•59-0•79]). The most commonly assigned COMCATs by InterVA-5 were emergencies (68 [48%] of 143 deaths) and health systems (62 [43%] deaths), and by physician review (India only) were health systems (seven [44%] of 16 deaths) and inevitability (five [31%] deaths); agreement between InterVA-5 and physician review (India data only) was poor (κ=0•04 [0•00-0•15]). Interpretation Our findings indicate that InterVA-5 is less accurate than InterVA-4 at ascertaining causes and circumstances of maternal death, when compared with physician review. Our results suggest a need to improve the next iteration of InterVA, and for researchers and clinicians to preferentially use InterVA-4 when recording maternal deaths. Funding University of British Columbia (grantee of the Bill & Melinda Gates Foundation).
Telehealth is a strategy to expand the reach of pulmonary rehabilitation (PR). Smartphones can mo... more Telehealth is a strategy to expand the reach of pulmonary rehabilitation (PR). Smartphones can monitor and transmit oxygen saturation (SpO 2 ) and heart rate (HR) data to ensure patient safety during home-based or other exercise. The purpose of this study was to evaluate the usability, validity and reliability of a Kenek O 2 pulse oximeter and custom prototype smartphone application (smartphone oximeter) during rest and exercise in healthy participants and those with chronic lung disease. Methods Fifteen individuals with chronic lung disease and 15 healthy controls were recruited. SpO 2 and HR were evaluated at rest and during cycling and walking. SpO2 was valid if the mean bias was within +± 2%, the level of agreement (LoA) was within ± 4%; HR was valid if the mean bias was within ± 5 beats per min (bpm), LoA was within ± 10 bpm. Usability was assessed with a questionnaire and direct observation. The smartphone oximeter was deemed easy to use. At rest, SpO 2 measures were valid in both groups (bias <2%, lower bound LoA -2 to 3%). During exercise, SpO 2 measurement did not meet validity and reliability thresholds in the patients with chronic lung disease, but was accurate for the healthy controls. HR recording during exercise or rest was not valid (LoA > 10 bpm) in either group. Conclusions The smartphone oximeter did not record HR or SpO 2 accurately in patients with chronic lung disease during exercise, although SpO 2 was valid at rest. During exercise, patients with chronic lung disease should pause to ensure greatest accuracy of SpO 2 and HR measurement.
2022 44th Annual International Conference of the IEEE Engineering in Medicine & Biology Society (EMBC), Jul 11, 2022
Neonatal respiratory distress is a common condition that if left untreated, can lead to short-and... more Neonatal respiratory distress is a common condition that if left untreated, can lead to short-and long-term complications. This paper investigates the usage of digital stethoscope recorded chest sounds taken within 1 min post-delivery, to enable early detection and prediction of neonatal respiratory distress. Fifty-one term newborns were included in this study, 9 of whom developed respiratory distress. For each newborn, 1 min anterior and posterior recordings were taken. These recordings were pre-processed to remove noisy segments and obtain high-quality heart and lung sounds. The random undersampling boosting (RUSBoost) classifier was then trained on a variety of features, such as power and vital sign features extracted from the heart and lung sounds. The RUSBoost algorithm produced specificity, sensitivity, and accuracy results of 85.0%, 66.7% and 81.8%, respectively. Clinical relevance-This paper investigates the feasibility of digital stethoscope recorded chest sounds for early detection of respiratory distress in term newborn babies, to enable timely treatment and management. E.
In response to: Spurr R, Ng E, Onchiri FM, Rapha B, Nakatumba-Nabende J, Rosenfeld M, Najjingo I,... more In response to: Spurr R, Ng E, Onchiri FM, Rapha B, Nakatumba-Nabende J, Rosenfeld M, Najjingo I, Stout J, Nantanda R, Ellington LE. Performance and usability of a new mobile application for measuring respiratory rate in young children with acute lower respiratory infections. Pediatr Pulmonol. 2022 Aug 22. doi: 10.1002/ppul.26125.
IEEE Transactions on Biomedical Engineering, Oct 1, 2019
The goal of this study was to optimize robust PID control for propofol anesthesia in children age... more The goal of this study was to optimize robust PID control for propofol anesthesia in children aged 5-10 years to improve performance, particularly to decrease the time of induction of anesthesia while maintaining robustness. Methods: We analyzed results of a previous study conducted by our group to identify opportunities for system improvement. Allometric scaling was introduced to reduce the interpatient variability and a new robust PID controller was designed using an optimization based method. We evaluated this optimized design in a clinical study involving 16 new cases. Results: The optimized controller design achieved the performance predicted in simulation studies in the design stage. Time of induction of anesthesia was median [Q1, Q3] 3.7 [2.3, 4.1] minutes and the achieved global score was 13.4 [9.9, 16.8]. Conclusion: Allometric scaling reduces the interpatient variability in this age group, and allows for improved closed-loop performance. The uncertainty described by the model set, the predicted closedloop responses and the predicted robustness margins are realistic. The system meets the design objectives of improved speed of induction of anesthesia while maintaining robustness, improving clinically relevant system behavior. Significance: Control system optimization and ongoing system improvement are essential to the development of a clinically relevant commercial device. This paper demonstrates the validity of our approach, including system modeling, controller optimization and pre-clinical testing in simulation.
Uploads
Papers by Guy Dumont