Continuous and non-invasive monitoring of dengue patients has the potential to improve care and outcomes. Our investigation leverages photoplethysmography (PPG) to design a system capable of assessing patients’ progression throughout dengue fever and assist clinicians in delivering higher-quality care.

The thesis starts by developing a novel PPG acquisition system capable of multi-site and multi-wavelength operation. We demonstrate the ability of the novel system to acquire PPG across 5 different wavelengths, from fingertip and wrist at a sampling rate of 1,000Hz. Furthermore, by evaluating the frequency spectrum and calculating ratio-based signal measures between different wavelengths we confirm its sensitivity to heart rate, oxygen saturation and haematocrit. To overcome negative effects on waveform morphology due to the high-frequency multi-wavelength nature of the signal we propose and validate compensatory algorithms.

The performance of the system was evaluated on dengue patients in two clinical investigations, one using a commercial pulse oximeter as a baseline and the other using the developed system. PPG data from 238 patients were analyzed, with 50 of them having multiwavelength data. The system demonstrated the capability of absolute measurement and sensitivity to changes in haematocrit, achieving a correlation of over 0.7 between PPG signal data and changes in invasively measured haematocrit in 83% of suitable patients. The acquired data was also investigated by convolutional neural networks (CNNs) to assess their relationship to dengue. Using the multi-wavelength data, an accuracy of 0.76 between severe and uncomplicated patients was achieved, showing a 7.74% improvement in dengue severity classification compared to single- wavelength approach.

The work concludes with an outline of a regulatory framework in the form of user needs as defined in the internationally recognized ISO 13485 medical device standard to bring the wearable system to market in relevant low- and middle-income (LMIC) regions. Overall, the investigation demonstrates the effectiveness of a novel PPG acquisition system for continuous and non-invasive monitoring of dengue patients and opens avenues for translation with the potential to lead to improved patient outcomes.