Disease surveillance is used to monitor ongoing control activities, detect early outbreaks, and inform intervention priorities and policies. However, data from disease surveillance that could be used to support real-time decisionmaking remain largely underutilised. Using the Brazilian Amazon malaria surveillance dataset as a case study, in this paper we explore the potential for unsupervised anomaly detection machine learning techniques to discover signals of epidemiological interest. We found that our models were able to provide an early indication of outbreak onset, outbreak peaks, and change points in the proportion of positive malaria cases. Specifically, the sustained rise in malaria in the Brazilian Amazon in 2016 was flagged by several models. We found that no single model detected all anomalies across all health regions. Because of this, we provide the minimum number of machine learning models top-k models) to maximise the number of anomalies detected across different health regions. We discovered that the top three models that maximise the coverage of the number and types of anomalies detected across the thirteen health regions are principal component analysis, stochastic outlier selection, and the minimum covariance determinant. Anomaly detection is a potentially valuable approach to discovering patterns of epidemiological importance when confronted with a large volume of data across space and time. Our exploratory approach can be replicated for other diseases and locations to inform monitoring, timely interventions, and actions towards the goal of controlling endemic disease.

The success of deep machine learning (DML) models in gaming and robotics has increased its trial in clinical and public healthcare solutions. In applying DML to healthcare problems, a special challenge of inadequate electrical energy and computing resources exists in regional and developing areas of the world. In this paper, we evaluate and report the computational and predictive performance design trade-offs for four candidate deep learning models that can be deployed for rapid malaria case finding. The goal is to maximise malaria detection accuracy while reducing computing resource and energy consumption. Based on our experimental results using a blood smear malaria test data set, the quantised versions of Basic Convolutional Neural Network (B-CNN) and MobileNetV2 have better malaria detection performance (up to 99% recall), lower memory usage (2MB 8-bit quantised model) and shorter inference time (33–95 microseconds on mobile phones) than VGG-19 fine-tuned and quantised models. Hence, we have implemented MobileNetV2 in our mobile application as it has even a lower memory requirement than B-CNN. This work will help to counter the negative effects of COVID-19 on the previous successes towards global malaria elimination.

Medical image scans and associated electronic medical records (EMR) could be stored locally or transmitted for use in autodiagnosis and remote healthcare in teleradiology. Hence, they require security against unauthorised access and modification. Among other means of providing this security, information hiding (IH) techniques have gained relevance especially for open networks that are prone to active attacks. However, the evaluation of the suitability of these IH algorithms in terms of preserving medical image diagnostic features is currently limited to signal processing parameters. This paper re-interprets existing evaluation parameters and provides a new framework that allows dynamic selection of medical image IH (watermarking and steganography) security algorithms. Specifically, criteria that capture medical statistics used in the diagnosis and monitoring of patients were incorporated. These criteria and framework were validated on the Pneumonia Chest Xray dataset (used in a Kaggle Competition) using three selected IH algorithms that offer privacy and image tamper detection.

This paper presents a capacity-improved Spread Spectrum (SS) watermarking method for data hiding and security for medical image and Electronic Medical Record (EMR) transmission in Teleradiology. SS watermarking is more secure than other watermarking methods but currently has low data carrying capacity due to the spreading of a single bit in larger cover sample in order to achieve higher robustness and security. In this work, a new method in spatial domain is proposed in order to improve on the current hiding capacity of just one bit per sample to up to 6 bits per sample (8×8 pixel block). This new watermark compression encoding method is suitably adapted for blind SS watermarking for hiding patient data securely in medical images of high pixel depth. The new method was successfully implemented and evaluated using Magnetic Resonance Image (MRI) scans.

Visible light photography diagnostic images are coloured ex vivo medical images popularly used in Dermatology and Endoscopy for diagnosis and monitoring. The need to protect the integrity of these images as well as associated patient data calls for techniques such as image steganography and watermarking. This research explores and compares the effect of watermarking on the YIQ and YCbCr colour transforms used in processing digital coloured images and video in recent times. Using a new spread spectrum watermarking algorithm, it was found that YIQ has better distortion performance than YCbCr in the order of 3dB while YCbCr had lower BER for accurate watermark retrieval and tamper detection in the order of 1.3 x 10-3.