Severity detection tool for patients with infectious disease

Abstract

Hand foot and mouth disease (HFMD) and tetanus are serious infectious diseases in low and middle income countries. Tetanus in particular has a high mortality rate and its treatment is resource-demanding. Furthermore, HFMD often affects a large number of infants and young children. As a result, its treatment consumes enormous healthcare resources, especially when outbreaks occur. Autonomic nervous system dysfunction (ANSD) is the main cause of death for both HFMD and tetanus patients. However, early detection of ANSD is a difficult and challenging problem. In this paper, we aim to provide a proof-of-principle to detect the ANSD level automatically by applying machine learning techniques to physiological patient data, such as electrocardiogram (ECG) and photoplethysmogram (PPG) waveforms, which can be collected using low-cost wearable sensors. Efficient features are extracted that encode variations in the waveforms in the time and frequency domains. A support vector machine is employed to classify the ANSD levels. The proposed approach is validated on multiple datasets of HFMD and tetanus patients in Vietnam. Results show that encouraging performance is achieved in classifying ANSD levels. Moreover, the proposed features are simple, more generalisable and outperformed the standard heart rate variability (HRV) analysis. The proposed approach would facilitate both the diagnosis and treatment of infectious diseases in low and middle income countries, and thereby improve overall patient care.

Multi-Modal Diagnosis of Infectious Diseases in the Developing World

“…Baseline -SVM performance (%) Method A P R S F1 PPG- [10] 70.2 ± 1.0 70.4 ± 0.8 92.6 ± 0.3 29.5 ± 2.9 80.0 ± 0.5 ECG- [10] 80.2 ± 0.7 78.4 ± 0.9 95.3 ± 0.5 53.4 ± 2.5 86.0 ± 0.4 FC- [10] 78. Table II shows the performance of the proposed approach for the on-site triage process compared with the baseline methods.…”

“…Baseline -SVM performance (%) Method A P R S F1 ECG- [8] 57.1 ± 0.2 35.0 ± 0.2 35.2 ± 0.2 67.6 ± 0.1 34.6 ± 0.2 ECG-2- [10] 70.9 ± 0.1 60.6 ± 0.1 55.9 ± 0.2 78.0 ± 0.1 55.7 ± 0.2 FC- [10] 70.2 ± 0.1 60.0 ± 0.1 54.5 ± 0. c) LR vs. SVM: LR performs competitively with SVM, and even exploits the feature fusion better than the SVM, i.e. 95.7% vs. 94.2%, respectively.…”

“…We compared the proposed approach with two existing works: our previous work [10] and a baseline method [8], which employed handcrafted features for the classification of severity levels. Abebe et al [10] applied simple time-and frequency-domain features whereas Malik et al [8] strictly required the detection of each QRS complex in ECG waveforms followed by the extraction of vital signs such as heart rate and RR intervals. We employed an SVM with a Gaussian kernel to validate the baseline features.…”

“…In previous work, we demonstrated the utility of the PPG and ECG physiological signals at predicting ANSD severity levels arising from tetanus and HFMD infections, using handcrafted features [10], [18]. In this paper we leverage multimodal ML and transfer learning to produce deep learning architectures to better predict infection severity using signals that could be obtained from inexpensive wearable sensors.…”

“…Moreover, QRS detection incurs additional computational cost. Previously, we have demonstrated that generic time and frequency features [10] outperformed the traditional heart rate variability (HRV) features [8] across multiple datasets of infectious disease patients. However, the diagnosis performance still has room for improvement, particularly using latest modeling techniques (e.g.…”