SepAl: Sepsis Alerts On Low Power Wearables With Digital Biomarkers and On-Device Tiny Machine Learning

TL;DR

SepAl demonstrates an energy-efficient, on-device sepsis alert system that operates on low-power wearables by extracting six digital vital signs and running a tiny temporal convolutional neural network. The method achieves real-time detection through a per-vital-sign TCN ensemble and a consensus-based alert mechanism, with quantisation enabling deployment on Cortex-M33 microcontrollers. Retrospective analysis yields strong discrimination (e.g., ~0.83 sensitivity and ~0.79 specificity at 4 hours), while real-time evaluation reveals trade-offs between stride, aggregation, and timely alerts, achieving a median time-to-sepsis near $9.8$ hours. The INT8 quantised model maintains close accuracy to the floating-point version, while delivering low latency ($\ ext{latency} \approx 143$ ms) and low energy consumption ($\approx 2.68$ mJ per inference), highlighting the practicality of edge AI for wearable sepsis monitoring. The work contributes a concrete pathway for deployable, privacy-preserving, multi-vital wearable surveillance with open-source code for broader adoption.

Abstract

Sepsis is a lethal syndrome of organ dysfunction that is triggered by an infection and claims 11 million lives per year globally. Prognostic algorithms based on deep learning have shown promise in detecting the onset of sepsis hours before the actual event but use a large number of bio-markers, including vital signs and laboratory tests. The latter makes the deployment of such systems outside hospitals or in resource-limited environments extremely challenging. This paper introduces SepAl, an energy-efficient and lightweight neural network, using only data from low-power wearable sensors, such as photoplethysmography (PPG), inertial measurement units (IMU), and body temperature sensors, designed to deliver alerts in real-time. SepAl leverages only six digitally acquirable vital signs and tiny machine learning algorithms, enabling on-device real-time sepsis prediction. SepAl uses a lightweight temporal convolution neural network capable of providing sepsis alerts with a median predicted time to sepsis of 9.8 hours. The model has been fully quantized, being able to be deployed on any low-power processors, and evaluated on an ARM Cortex-M33 core. Experimental evaluations show an inference efficiency of 0.11MAC/Cycle and a latency of 143ms, with an energy per inference of 2.68mJ. This work aims at paving the way toward accurate disease prediction, deployable in a long-lasting multi-vital sign wearable device, suitable for providing sepsis onset alerts at the point of care. The code used in this work has been open-sourced and is available at https://github.com/mgiordy/sepsis-prediction

Figures (6)

• Figure 1:
• Figure 2:
• Figure 4: Data processing pipeline from sensor data acquisition to on-board vital sign extraction, which consists of multiple digital signal processing steps. After vital sign extraction, the vitals are fed to the neural network and then to the consensus algorithm for the prediction of sepsis onset. (a) Multi-modal temporal convolution neural network architecture. (b) Data windows organization for the retrospective analysis. (c) Data windows organization and labeling for neural network real-time training. (c) Data windows prediction and consensus algorithm used for model evaluation and validation in real-time.
• Figure 5: Retrospective analysis over different prediction times.
• Figure 6: Parameter study on the online model varying the windows strides and the aggregation parameter.