Search for: All records

Epilepsy is one of the most common neurological diseases globally (around 50M people globally). Fortunately, up to 70% of people with epilepsy could live seizure-free if properly diagnosed and treated, and a reliable technique to monitor the onset of seizures could improve the quality of life of patients who are constantly facing the fear of random seizure attacks. The current gold standard, video-EEG (v-EEG), involves attaching over 20 electrodes to the scalp, is costly, requires hospitalization, trained professionals, and is uncomfortable for patients. To address this gap, we developedEarSD, a lightweight and unobtrusive ear-worn system to detect seizure onsets by measuring physiological signals behind the ears. This system can be integrated into earphones, headphones, or hearing aids, providing a convenient solution for continuous monitoring.EarSDis an integrated custom-builtsensing-computing-communicationear-worn platform to capture seizure signals, remove the noises caused by motion artifacts and environmental impacts, and stream the collected data wirelessly to the computer/mobile phone nearby.EarSD’s ML algorithm, running on a server, identifies seizure-associated signatures and detects onset events. We evaluated the proposed system in both in-lab and in-hospital experiments at the University of Texas Southwestern Medical Center with epileptic seizure patients, confirming its usability and practicality. 

Traditional machine learning techniques are prone to generating inaccurate predictions when confronted with shifts in the distribution of data between the training and testing phases. This vulnerability can lead to severe consequences, especially in applications such as mobile healthcare. Uncertainty estimation has the potential to mitigate this issue by assessing the reliability of a model's output. However, existing uncertainty estimation techniques often require substantial computational resources and memory, making them impractical for implementation on microcontrollers (MCUs). This limitation hinders the feasibility of many important on-device wearable event detection (WED) applications, such as heart attack detection. In this paper, we present UR2M, a novel Uncertainty and Resource-aware event detection framework for MCUs. Specifically, we (i) develop an uncertainty-aware WED based on evidential theory for accurate event detection and reliable uncertainty estimation; (ii) introduce a cascade ML framework to achieve efficient model inference via early exits, by sharing shallower model layers among different event models; (iii) optimize the deployment of the model and MCU library for system efficiency. We conducted extensive experiments and compared UR2M to traditional uncertainty baselines using three wearable datasets. Our results demonstrate that UR2M achieves up to 864% faster inference speed, 857% energy-saving for uncertainty estimation, 55% memory saving on two popular MCUs, and a 22% improvement in uncertainty quantification performance. UR2M can be deployed on a wide range of MCUs, significantly expanding real-time and reliable WED applications. 

Difficulty falling asleep is one of the typical insomnia symptoms. However, intervention therapies available nowadays, ranging from pharmaceutical to hi-tech tailored solutions, remain ineffective due to their lack of precise real-time sleep tracking, in-time feedback on the therapies, and an ability to keep people asleep during the night. This paper aims to enhance the efficacy of such an intervention by proposing a novel sleep aid system that can sense multiple physiological signals continuously and simultaneously control auditory stimulation to evoke appropriate brain responses for fast sleep promotion. The system, a lightweight, comfortable, and user-friendly headband, employs a comprehensive set of algorithms and dedicated own-designed audio stimuli. Compared to the gold-standard device in 883 sleep studies on 377 subjects, the proposed system achieves (1) a strong correlation (0.89 ± 0.03) between the physiological signals acquired by ours and those from the gold-standard PSG, (2) an 87.8% agreement on automatic sleep scoring with the consensus scored by sleep technicians, and (3) a successful non-pharmacological real-time stimulation to shorten the duration of sleep falling by 24.1 min. Conclusively, our solution exceeds existing ones in promoting fast falling asleep, tracking sleep state accurately, and achieving high social acceptance through a reliable large-scale evaluation. 

While the global healthcare market of wearable devices has been growing significantly in recent years and is predicted to reach $60 billion by 2028, many important healthcare applications such as seizure monitoring, drowsiness detection, etc. have not been deployed due to the limited battery lifetime, slow response rate, and inadequate biosignal quality.This study proposes PROS, an efficient pattern-driven compressive sensing framework for low-power biopotential-based wearables. PROS eliminates the conventional trade-off between signal quality, response time, and power consumption by introducing tiny pattern recognition primitives and a pattern-driven compressive sensing technique that exploits the sparsity of biosignals. Specifically, we (i) develop tiny machine learning models to eliminate irrelevant biosignal patterns, (ii) efficiently perform compressive sampling of relevant biosignals with appropriate sparse wavelet domains, and (iii) optimize hardware and OS operations to push processing efficiency. PROS also provides an abstraction layer, so the application only needs to care about detected relevant biosignal patterns without knowing the optimizations underneath.We have implemented and evaluated PROS on two open biosignal datasets with 120 subjects and six biosignal patterns. The experimental results on unknown subjects of a practical use case such as epileptic seizure monitoring are very encouraging. PROS can reduce the streaming data rate by 24X while maintaining high fidelity signal. It boosts the power efficiency of the wearable device by more than 1200\% and enables the ability to react to critical events immediately on the device. The memory and runtime overheads of PROS are minimal, with a few KBs and 10s of milliseconds for each biosignal pattern, respectively. PROS is currently adopted in research projects in multiple universities and hospitals. 

Frequent blood pressure monitoring is the key to diagnosis and treatments of many severe diseases. However, the conventional ambulatory methods require patients to carry a blood pressure (BP) monitoring device for 24 h and conduct the measurement every 10--15 min. Despite their extensive usage, wearing the wrist/arm-based BP monitoring device for a long time has a significant impact on users' daily activities. To address the problem, we developed eBP to measure blood pressure (BP) from inside user's ear aiming to minimize the measurement's impact on users' normal activities although maximizing its comfort level. The key novelty of eBP includes (1) a light-based inflatable pulse sensor which goes inside the ear, (2) a digital air pump with a fine controller, and (3) BP estimation algorithms that eliminate the need of blocking the blood flow inside the ear. Through the comparative study of 35 subjects, eBP can achieve the average error of 1.8 mmHg for systolic (high-pressure value) and -3.1 mmHg for diastolic (low-pressure value) with the standard deviation error of 7.2 mmHg and 7.9 mmHg, respectively. These results satisfy the FDA's AAMI standard, which requires a mean error of less than 5 mmHg and a standard deviation of less than 8 mmHg. 

Diagnosing hypertension or hemodialysis requires patients to carry a blood pressure (BP) monitoring device for 24 hours. Th erefore, wearing the wrist/arm-based BP monitoring device, in this case, has a signifi cant impact on users' daily activities. To address the problem, we developed eBP, an ear-worn device that measures blood pressure from inside the ear. Th rough the evaluation of 35 subjects, eBP can achieve the average error of 1.8 mmHg for systolic BP and -3.1 mmHg for diastolic BP with the standard deviation error of 7.2 mmHg and 7.9 mmHg, respectively.