10:30
Monitoring
Chair: Pieter van Gorp
QUICC: Quantification of Urgent Indicators for early detection of Complications after Colorectal resection; Proof of principle study on early detection of postoperative complications using continuous heart rate and respiratory rate measurements, in bot
Ilse Waanders, Harry Vaassen, Daan Lips, Arlene John, Annemieke Witteveen
Abstract: Rationale: Postoperative complications after colorectal surgery, such as anastomotic leakage (AL), significantly increase patient morbidity and mortality. Early detection and treatment are crucial for improving patient outcomes. Continuous vital sign monitoring enables earlier alarming based on changes in patient vital signs. This study investigates the potential of continuous heart rate (HR) and respiratory rate (RR) monitoring, to facilitate early detection of AL without increasing alarm fatigue. Methods: A retrospective analysis was conducted on continuous HR and RR measured with the Philips HealthDot, collected over 14 days postoperatively, from patients who underwent elective oncological colorectal surgery. HR and RR from patients with AL were compared to those without complications. A novel trend analysis of the continuous HR is compared to the golden standard a threshold-based alarming; the Remote Early Warning Score (REWS). An alarm is generated in the novel trend-based method when the moving average of 24 hours of HR data is larger than 8 beats per minute over 4 hours and the derivative is positive. The two methods were compared based on the numbers of false positives (FP) in patients without complications and time of first alarm in hours before CT-scan. Results: The REWS is compared to the trend alarming of data from two patients with AL and six patients without complications. The REWS methods alarmed 72 and 8 hours prior to diagnosis with two false positive alarms, while the trend-based strategy alarmed 87 and 10 hours before diagnosis with no false positives. In one patient, both methods alarmed prior to onset of postoperative complication symptoms. Conclusions: This small proof-of-principle study demonstrates the potential of continuous vital sign monitoring for early detection of AL after colorectal resection, both in-hospital and at home. Trend-based method alarmed earlier than REWS, without false positives. Although limited by the small sample size, these findings support further investigation into personalized alarms based on trend analysis of continuously measured HR as an effective method for early detection of AL after colorectal resections.
|
Measurement of sweat gland activity by sweat sensing and deep learning
Jelte Haakma, Elisabetta Peri, Simona Turco
Abstract: Abstract—Semi-continuous monitoring of general ward patients is expected to reduce
hospital mortality. Sweat is a biomarker-rich biofluid with potential for non-invasive, semicontinuous patient monitoring. Efforts to correlate the biomarkers in sweat to the condition of
a patient have yielded only limited results. The only clinically approved use of sweat
biomarkers is the diagnosis of cystic fibrosis. Knowledge of the sweat rate per gland and the
number of active sweat glands from which the sweat is sampled is difficult to obtain and yet
plays an important role to accurately estimate the concentration of biomarkers found in sweat.
To estimate the sweat rate per sweat gland a discrete sweat-sensing device can be used. A
model was created to generate synthetic signals of such a discrete sweat-sensing device. In
this work, we adopt a deep-learning strategy to estimate the sweat rate per gland, and the
number of active sweat glands, based on the simulated signals, with the aim of demonstrating
the feasibility of this approach. Our approach demonstrated its capability to estimate the sweat
rate within a 10% margin of error across all tested datasets. The number of glands could be
estimated with a minimum accuracy of 70%. This shows that deep learning is a promising
method to interpret the signals of a discrete sweat-sensing device.
|
Change Point Detection for Continuous Physiological Monitoring using Wearables
Welmoed Tjepkema, Arlene John, Lisette Vernooij, Champika Ranasinghe, Bert-Jan van Beijnum, Martine Breteler
Abstract: Background: Continuous patient monitoring using wearable devices can improve patient outcomes by enabling earlier detection of deterioration in general wards. However, current threshold-based alarms produce excessive false alarms, causing alarm fatigue. As monitored patient numbers grow, healthcare professionals struggle to interpret vital sign trends and prioritize patients effectively. This research explores utilising an online, causal unsupervised multivariate change point detection model to continuously measured vital signs with a wearable multiparameter system, that measures heart rate, respiratory rate, temperature, blood pressure, oxygen saturation, and activity, at the UMC Utrecht to address these challenges.
Methods: Monitoring data of 24 hours from 52 selected patients, including eight patients with known events were used to validate an online unsupervised change point detection model: the adaptive LSTM-Autoencoder Change-Point Detection (ALACPD) model. Missing data were imputed causally using the Kalman filter. Change points for validation were annotated by six medical service centre employees, with inter-rater reliability assessed via the Fleiss’ kappa statistic. Precision, recall, F1-score, and covering score were used for the validation of three variants of the ALACPD model: AE-skipLSTM-AR, AE-skipLSTM, and AR.
Results: The observed agreement among the annotators was high, with pe = 0.98, indicating high inter-rater reliability for validation labels. The AE-skipLSTM-AR outperformed the other model variants and demonstrated moderate performance, achieving an F1-score of 0.51, a recall of 0.7, precision of 0.44, and covering score of 0.51. The model successfully detected all known events (e.g., ICU transfer or atrial fibrillation episodes) and generated an average of 6.1 alarms/patient/day. It was observed that the annotators and ALACPD model demonstrated high accuracy in identifying abrupt changes in the mean or variance of the vital sign data while variability was observed when identifying gradual changes.
Conclusions: The ALACPD model shows promise for continuous vital sign monitoring in ward patients to reduce alarm fatigue while still capturing key events. Additionally, the model can offer valuable insights in interpreting trends and can assist in prioritising patients in the future. Future research should focus on automated hyperparameter tuning, improving annotation reliability, and addressing missing data challenges to optimize the model’s performance and generalizability in clinical settings.
|
Influence of environment on wetting properties of common polymers for Skin-Device Interfaces
Hanneke Reuvekamp, Edsko Hekman, Emile van der Heide, Dave Matthews
Abstract: Many non-invasive personal healthcare devices suffer from poor, occlusive skin contact. This disrupts
physiological interface microclimate and device functionality. Surface engineering can provide
multifunctional surfaces which precisely control and regulate microclimates at the skin-device interface.
The regulation efficacy through surface engineering is contingent upon both the material and intrinsic
surface properties, as well as the topology design. In particular, the wettability plays a crucial role in
regulating these microclimate conditions. Before introducing surface engineering for developing
innovative skin-device interfaces, the wetting behaviour of the materials needs to be determined. The
polymeric materials polydimethylsiloxane (PDMS), polyvinyl chloride (PVC), polypropylene (PP), and
polyethylene (PE) are commonly used in non-invasive personal healthcare devices. Existing studies
indicate an impact of temperature and relative humidity (RH) on the wetting properties, as characterised
by the contact angle (CA). Nevertheless, most studies have been conducted under ambient conditions
(20°C and 50% RH), on different materials, or are in disagreement regarding the climate dependence of
polymer wettability. This study applies advanced characterisation and evaluation methods to set a
reference for the wetting behaviour of the mentioned polymers in their untreated state. A custom climate
chamber was designed and built in which temperature and humidity can be separately controlled. In this
chamber, droplets can be applied on a surface, after which the CA can be measured using a goniometer.
In this controlled environment, the CA of PP, PVC, PE and PDMS substrates was determined as a
function of relative humidity (RH 10 to 90%) and temperature (5 to 50°C). The systematic and
comprehensive study shows that although the CA is significantly different for almost all polymers, there
is no significant dependence of the CA on either humidity or temperature for PP, PVC and PE. The water
CA of PDMS exhibits a linear temperature dependency at a constant RH, while the CA measured with
diiodomethane suggests a linearly inverse dependence on RH. We conclude that the influence of these
climate conditions on the wettability of these polymers is negligible. For improvement of the device-
skin interface we will therefore focus on engineering surface engineering strategies for microclimate
regulation.Many non-invasive personal healthcare devices suffer from poor, occlusive skin contact. This disrupts
physiological interface microclimate and device functionality. Surface engineering can provide
multifunctional surfaces which precisely control and regulate microclimates at the skin-device interface.
The regulation efficacy through surface engineering is contingent upon both the material and intrinsic
surface properties, as well as the topology design. In particular, the wettability plays a crucial role in
regulating these microclimate conditions. Before introducing surface engineering for developing
innovative skin-device interfaces, the wetting behaviour of the materials needs to be determined. The
polymeric materials polydimethylsiloxane (PDMS), polyvinyl chloride (PVC), polypropylene (PP), and
polyethylene (PE) are commonly used in non-invasive personal healthcare devices. Existing studies
indicate an impact of temperature and relative humidity (RH) on the wetting properties, as characterised
by the contact angle (CA). Nevertheless, most studies have been conducted under ambient conditions
(20°C and 50% RH), on different materials, or are in disagreement regarding the climate dependence of
polymer wettability. This study applies advanced characterisation and evaluation methods to set a
reference for the wetting behaviour of the mentioned polymers in their untreated state. A custom climate
chamber was designed and built in which temperature and humidity can be separately controlled. In this
chamber, droplets can be applied on a surface, after which the CA can be measured using a goniometer.
In this controlled environment, the CA of PP, PVC, PE and PDMS substrates was determined as a
function of relative humidity (RH 10 to 90%) and temperature (5 to 50°C). The systematic and
comprehensive study shows that although the CA is significantly different for almost all polymers, there
is no significant dependence of the CA on either humidity or temperature for PP, PVC and PE. The water
CA of PDMS exhibits a linear temperature dependency at a constant RH, while the CA measured with
diiodomethane suggests a linearly inverse dependence on RH. We conclude that the influence of these
climate conditions on the wettability of these polymers is negligible. For improvement of the device-
skin interface we will therefore focus on engineering surface engineering strategies for microclimate
regulation.Many non-invasive personal healthcare devices suffer from poor, occlusive skin contact. This disrupts
physiological interface microclimate and device functionality. Surface engineering can provide
multifunctional surfaces which precisely control and regulate microclimates at the skin-device interface.
The regulation efficacy through surface engineering is contingent upon both the material and intrinsic
surface properties, as well as the topology design. In particular, the wettability plays a crucial role in
regulating these microclimate conditions. Before introducing surface engineering for developing
innovative skin-device interfaces, the wetting behaviour of the materials needs to be determined. The
polymeric materials polydimethylsiloxane (PDMS), polyvinyl chloride (PVC), polypropylene (PP), and
polyethylene (PE) are commonly used in non-invasive personal healthcare devices. Existing studies
indicate an impact of temperature and relative humidity (RH) on the wetting properties, as characterised
by the contact angle (CA). Nevertheless, most studies have been conducted under ambient conditions
(20°C and 50% RH), on different materials, or are in disagreement regarding the climate dependence of
polymer wettability. This study applies advanced characterisation and evaluation methods to set a
reference for the wetting behaviour of the mentioned polymers in their untreated state. A custom climate
chamber was designed and built in which temperature and humidity can be separately controlled. In this
chamber, droplets can be applied on a surface, after which the CA can be measured using a goniometer.
In this controlled environment, the CA of PP, PVC, PE and PDMS substrates was determined as a
function of relative humidity (RH 10 to 90%) and temperature (5 to 50°C). The systematic and
comprehensive study shows that although the CA is significantly different for almost all polymers, there
is no significant dependence of the CA on either humidity or temperature for PP, PVC and PE. The water
CA of PDMS exhibits a linear temperature dependency at a constant RH, while the CA measured with
diiodomethane suggests a linearly inverse dependence on RH. We conclude that the influence of these
climate conditions on the wettability of these polymers is negligible. For improvement of the device-
skin interface we will therefore focus on engineering surface engineering strategies for microclimate
regulation.Many non-invasive personal healthcare devices suffer from poor, occlusive skin contact. This disrupts
physiological interface microclimate and device functionality. Surface engineering can provide
multifunctional surfaces which precisely control and regulate microclimates at the skin-device interface.
The regulation efficacy through surface engineering is contingent upon both the material and intrinsic
surface properties, as well as the topology design. In particular, the wettability plays a crucial role in
regulating these microclimate conditions. Before introducing surface engineering for developing
innovative skin-device interfaces, the wetting behaviour of the materials needs to be determined. The
polymeric materials polydimethylsiloxane (PDMS), polyvinyl chloride (PVC), polypropylene (PP), and
polyethylene (PE) are commonly used in non-invasive personal healthcare devices. Existing studies
indicate an impact of temperature and relative humidity (RH) on the wetting properties, as characterised
by the contact angle (CA). Nevertheless, most studies have been conducted under ambient conditions
(20°C and 50% RH), on different materials, or are in disagreement regarding the climate dependence of
polymer wettability. This study applies advanced characterisation and evaluation methods to set a
reference for the wetting behaviour of the mentioned polymers in their untreated state. A custom climate
chamber was designed and built in which temperature and humidity can be separately controlled. In this
chamber, droplets can be applied on a surface, after which the CA can be measured using a goniometer.
In this controlled environment, the CA of PP, PVC, PE and PDMS substrates was determined as a
function of relative humidity (RH 10 to 90%) and temperature (5 to 50°C). The systematic and
comprehensive study shows that although the CA is significantly different for almost all polymers, there
is no significant dependence of the CA on either humidity or temperature for PP, PVC and PE. The water
CA of PDMS exhibits a linear temperature dependency at a constant RH, while the CA measured with
diiodomethane suggests a linearly inverse dependence on RH. We conclude that the influence of these
climate conditions on the wettability of these polymers is negligible. For improvement of the device-
skin interface we will therefore focus on engineering surface engineering strategies for microclimate
regulation.
|
A novel kinetic model predicting the urea concentration in plasma during non-invasive sweat-based monitoring in hemodialysis
Xiaoyu Yin, Sophie Adelaars, Elisabetta Peri, Eduard Pelssers, Jaap den Toonder, Arthur Bouwman, Daan van de Kerkhof, Massimo Mischi
Abstract: The effectiveness of hemodialysis in patients with end-stage renal disease is frequently evaluated by monitoring changes in blood urea concentrations before and after the procedure. As monitoring of urea concentrations typically requires blood sampling, the development of sweat-sensing technology offers a possible less-invasive alternative to repeated venipuncture. Moreover, sweat-based urea monitoring could enable personalized treatment in a home-based setting. However, the clinical interpretation of sweat monitoring is hampered by the unclear correlation between urea concentrations in sweat and blood, which is regulated by complex active and passive transport mechanisms across the sweat gland. This study introduces a novel approach to estimate blood urea concentrations using sweat urea concentration values as input.
To simulate the complex transport mechanisms, a novel pharmacokinetic urea transport model was proposed. Such transport model together with a double-loop optimization strategy from our previous work were employed for patient-specific estimation of blood urea concentration. 32 patient samples of paired sweat and blood urea concentrations, collected both before and after hemodialysis, were used to validate the model. This resulted in an excellent Pearson correlation coefficient (0.98, 95%CI: 0.95-0.99) and a clinically irrelevant bias (-0.181 mmol/L before and -0.005 mmol/L after hemodialysis). This model enabled the accurate estimation of blood urea concentrations from sweat measurements.
This approach is not limited to urea monitoring in chronic kidney disease patients, opening new avenues for future research on non-invasive monitoring of blood biomarker concentrations through sweat-sensing technology, potentially supporting the development of novel patient-friendly monitoring applications in clinical practice.
|
Correlation between Glucose Responses from Continuous Glucose Monitoring and Glycaemic Index/Load from Food Intake
Linda Ong, Ming Cao, G.J (Bart) Verkerke, André van Beek, Elisabeth Wilhelm
Abstract: The International Diabetes Federation predicted that one in eight adults will live with diabetes by 2045 [1]. High glycaemic index (GI) and load (GL) diets are among the main risk for diabetes type 2 [2]. GI and GL are usually recorded using food diaries which rely on self-report with limited reliability [3]. One approach to tackle this limitation is by using Continuous Glucose Monitoring (CGM) responses to predict GI\GL of standardized meals to facilitate food tracking [4]. However, models that can predict GI\GL for unconstrained conditions are limited and not widely investigated. Therefore, new variables for such models need to be explored. We investigated the correlation between features extracted from CGM and GI / GL from food intake in an unconstraint condition.
Forty-eight healthy participants (21 males , 27 females), with an average age of 28.2 ± 2.7 and BMI of 23.4 ± 3.1, participated in an observational study for seven consecutive days. Participants used CGM (FreestyleLibre 2, Abbot, USA) and registered their main meals (breakfast, lunch, or dinner) in a food diary app (Mijn Eetmeter, Voedingscentrum, the Netherlands). We calculated fourteen features from CGM in two to six hours windowing after eating, which include Area Under Curve (AUC), rise AUC, fall AUC, relative amplitude, minimum amplitude, rise-time, fall-time, rise-slope, fall-slope, ratio between up and fall slope, mean, Standard Deviation (SD), Mean Amplitude of Glycemic Excursion (MAGE) and Coefficient Variation (CV).We used Spearman correlation due to the non-normal distribution data.
There was a moderate correlation between GL and both relative amplitude and SD (rs = .41, p < .01, 4 hours window) based on 120 recorded main meals. Glucose variability of SD, MAGE, and CV showed a weak correlation to GI for two to four hour windowing (rs = .24 – .31, p < .01). There was a weak correlation of mean (rs = .29 – .38, p < .01), rise-slope (rs = .24 – .36, p < .01) and MAGE (rs = .24 – .35, p < .01) to GL which were independent of window sizes.
In conclusion, our correlation analysis revealed that CGM amplitude, SD, MAGE, CV, mean, and rise-slope from CGM are promising features for models that aim at predicting glycaemic index and load of food intake in an unconstraint condition.
|
|