Abstract:Birth Apshyxia (BA) is a severe condition characterized by insufficient supply of oxygen to a newborn during the delivery. BA is one of the primary causes of neonatal death in the world. Although there has been a decline in neonatal deaths over the past two decades, the developing world, particularly sub-Saharan Africa, continues to experience the highest under-five (<5) mortality rates. While evidence-based methods are commonly used to detect BA in African healthcare settings, they can be subject to physician errors or delays in diagnosis, preventing timely interventions. Centralized Machine Learning (ML) methods demonstrated good performance in early detection of BA but require sensitive health data to leave their premises before training, which does not guarantee privacy and security. Healthcare institutions are therefore reluctant to adopt such solutions in Africa. To address this challenge, we suggest a federated learning (FL)-based software architecture, a distributed learning method that prioritizes privacy and security by design. We have developed a user-friendly and cost-effective mobile application embedding the FL pipeline for early detection of BA. Our Federated SVM model outperformed centralized SVM pipelines and Neural Networks (NN)-based methods in the existing literature

What problem does this paper attempt to address?

This paper attempts to solve several key problems in the early detection of neonatal asphyxia (Neonatal Asphyxia, BA), especially in view of the challenges faced by developing countries such as Africa. Specifically, the paper mainly focuses on the following points: 1. **Improve the accuracy of early detection**: Neonatal asphyxia is one of the leading causes of neonatal death, especially in sub - Saharan Africa, where this problem is particularly severe. Timely detection and intervention can significantly improve the prognosis of infants. However, existing evidence - based methods may be affected by misdiagnosis or delayed diagnosis by doctors in practical applications, thus hindering timely treatment. 2. **Protect the privacy and security of medical data**: Traditional centralized machine - learning methods require the transmission of sensitive medical data to a central server for training, which may lead to privacy leakage and security risks. Therefore, many medical institutions are cautious about adopting these solutions. 3. **Computing environment with limited resources**: Many medical institutions in developing countries lack sufficient computing resources to train large - scale machine - learning models. In addition, some existing automated diagnostic tools (such as Ubenwa) have not been widely deployed in Africa yet. To solve these problems, the author proposes a software architecture based on federated learning (Federated Learning, FL) - HumekaFL. Through the distributed learning method, HumekaFL uses local data on mobile devices for model training under the premise of ensuring data privacy and security. Its main features include: - **Privacy protection**: Through federated learning, sensitive medical data does not need to leave its institution, thus ensuring the security and privacy of data. - **Computing resource optimization**: Use a smaller model and train it on mobile devices, reducing the demand for high - performance computing resources. - **User - friendliness**: A user - friendly mobile application has been developed, which can be operated without requiring users to have professional skills, and is suitable for medical staff, parents or caregivers. In general, HumekaFL aims to provide an affordable and efficient solution to help with the early detection of neonatal asphyxia, thereby improving the health of newborns, especially in developing countries with limited resources. ### Formula summary The formulas involved in this paper are mainly used to describe the training process of support vector machines (SVM), specifically as follows: 1. **Objective function**: \[ F(w)=\frac{\lambda}{2}\|w\|_{2}^{2}+\frac{1}{n}\sum_{i = 1}^{n}l(w^{T}x_{i},y_{i}) \] where \(\lambda\) is the regularization parameter, and \(l(w^{T}x_{i},y_{i})=\max(0,1 - w^{T}x_{i}y_{i})\) is the Hinge loss function. 2. **Optimization problem**: \[ w^{*}=\arg\min_{w}F(w) \] 3. **Global model update**: \[ w_{t + 1}^{\text{global}}=\sum_{k\in S_{t}}\frac{n_{k}}{n}w_{t + 1}^{k} \] where \(S_{t}\) represents the set of randomly selected clients in the \(t\)-th round of communication, \(n_{k}\) is the amount of data of the \(k\)-th client, and \(n\) is the total amount of data. Through these methods, HumekaFL achieves efficient and secure early detection of neonatal asphyxia.

HumekaFL: Automated Detection of Neonatal Asphyxia Using Federated Learning

Democratizing AI in Africa: FL for Low-Resource Edge Devices

Harnessing Infant Cry for swift, cost-effective Diagnosis of Perinatal Asphyxia in low-resource settings

Ubenwa: Cry-based Diagnosis of Birth Asphyxia

An analysis of decipherable red blood cell abnormality detection under federated environment leveraging XAI incorporated deep learning

Federated Uncertainty-Aware Learning for Distributed Hospital EHR Data

Hierarchical Federated Learning based Anomaly Detection using Digital Twins for Smart Healthcare

Federated Machine Learning Based Fetal Health Prediction Empowered with Bio-Signal Cardiotocography

Feasibility Analysis of Federated Neural Networks for Explainable Detection of Atrial Fibrillation

Federated Anomaly Detection for Early-Stage Diagnosis of Autism Spectrum Disorders using Serious Game Data

Secur-e-Health Project: Towards Federated Learning for Smart Pediatric Care

FLICU: A Federated Learning Workflow for Intensive Care Unit Mortality Prediction

Autonomous fetal morphology scan: deep learning + clustering merger - the second pair of eyes behind the doctor

Federated Learning for Healthcare: Systematic Review and Architecture Proposal

A Fog-Based Privacy-Preserving Federated Learning System for Smart Healthcare Applications

A Federated Learning Approach to Breast Cancer Prediction in a Collaborative Learning Framework

Detecting Data Poisoning Attacks in Federated Learning for Healthcare Applications Using Deep Learning

Secure federated learning for Alzheimer's disease detection

Multimodal Federated Learning in Healthcare: a Review

Neonatal Disease Prediction Using Machine Learning Techniques

Federated learning aided breast cancer detection with intelligent Heuristic-based deep learning framework