Developers proposing new machine learning for health (ML4H) tools often pledge to match or even surpass the performance of existing tools, yet the reality is usually more complicated. Reliable deployment of ML4H to the real world is challenging as examples from diabetic retinopathy or Covid-19 screening show. We envision an integrated framework of algorithm auditing and quality control that provides a path towards the effective and reliable application of ML systems in healthcare. In this editorial, we give a summary of ongoing work towards that vision and announce a call for participation to the special issue Machine Learning for Health: Algorithm Auditing & Quality Control in this journal to advance the practice of ML4H auditing.

Digital health technologies can help tackle challenges in global public health. Digital and AI-for-Health Challenges, controlled events whose goal is to generate solutions to a given problem in a defined period of time, are one way of catalysing innovation. This article proposes an expanded investment framework for Global Health AI and digitalhealth Innovation that goes beyond traditional factors such as return on investment. Instead, we propose non monetary and non GDP metrics, such as Disability Adjusted Life Years or achievement of universal health coverage. Furthermore, we suggest a venture building approach around global health, which includes filtering of participants to reduce opportunity cost, close integration of implementation scientists and an incubator for the long-term development of ideas resulting from the challenge. Finally, we emphasize the need to strengthen human capital across a range of areas in local innovation, implementation-science, and in health services.

In this paper, we present a procedure for implementing field-based high-throughput plant phenotyping (HTPP) that can be used in resource-constrained research programs. The procedure relies on opensource tools with the only expensive item being one-off purchase of a drone. It includes acquiring images of the field of interest, stitching the images to get the entire field in one image, calculating and extracting the vegetation indices of the individual plots, and analyzing the extracted indices according to the experimental design. Two populations of groundnut genotypes with different maturities were evaluated for their reaction to early and late leaf spot (ELS, LLS) diseases under field conditions in 2020 and 2021. Each population was made up of 12 genotypes in 2020 and 18 genotypes in 2021. Evaluation of the genotypes was done in four locations in each year. We observed a strong correlation between the vegetation indices and the area under the disease progress curve (AUDPC) for ELS and LLS. However, the strength and direction of the correlation depended upon the time of disease onset, level of tolerance among the genotypes and the physiological traits the vegetation indices were associated with. In 2020, when the disease was observed to have set in late in medium duration population, at the beginning of the seed stage (R5), normalized green-red difference index (NGRDI) and variable atmospheric resistance index (VARI) derived at the beginning pod stage (R3) had a positive relationship with the AUDPC for ELS, and LLS. On the other hand, NGRDI and VARI derived from images taken at R5, and physiological maturity (R7) had negative relationships with AUDPC for ELS, and LLS. In 2021, when the disease was observed to have set in early (at R3) also in medium duration population, a negative relationship was observed between NGRDI and VARI and AUDPC for ELS and LLS, respectively. We found consistently negative relationships of NGRDI and VARI with AUDPC for ELS and LLS, respectively, within the short duration population in both years. Canopy cover (CaC), green area (GA), and greener area (GGA) only showed negative relationships with AUDPC for ELS and LLS when the disease caused yellowing and defoliation. The rankings of some genotypes changed for NGRDI, VARI, CaC, GA, GGA, and crop senescence index (CSI) when lesions caused by the infections of ELS and LLS became severe, although that did not affect groupings of genotypes when analyzed with principal component analysis. Notwithstanding, genotypes that consistently performed well across various reproductive stages with respect to the vegetation indices constituted the top performers when ELS, LLS, haulm, and pod yields were jointly considered.

A rapid and accurate diagnosis of medical conditions like cardiomegaly and pleural effusion is of the utmost importance to reduce mortality and medical costs. Artificial Intelligence has shown promise in diagnosing medical conditions. With this study, we seek to evaluate how well Artificial Intelligence (AI) systems, developed my minoHealth AI Labs, will perform at diagnosing cardiomegaly and pleural effusion, using chest x-rays from Ghana, Vietnam and the USA, and how well AI systems will perform when compared with radiologists working in Ghana. The evaluation dataset used in this study contained 100 images randomly selected from three datasets. The Deep Learning models were further tested on a larger Ghanaian dataset containing five hundred and sixty one (561) samples. Two AI systems were then evaluated on the evaluation dataset, whilst we also gave the same chest x-ray images within the evaluation dataset to 4 radiologists, with 5–20 years experience, to diagnose independently. For cardiomegaly, minoHealth.ai systems scored Area under the Receiver operating characteristic Curve (AUC-ROC) of 0.9 and 0.97 while the AUC-ROC of individual radiologists ranged from 0.77 to 0.87. For pleural effusion, the minoHealth.ai systems scored 0.97 and 0.91 whereas individual radiologists scored between 0.75 and 0.86. On both conditions, the best performing AI model outperforms the best performing radiologist by about 10%. We also evaluate the specificity, sensitivity, negative predictive value (NPV), and positive predictive value (PPV) between the minoHealth.ai systems and radiologists.

Purpose: Computer vision for automated analysis of cells and tissues usually include extracting features from images before analyzing such features via various machine learning and machine vision algorithms. The purpose of this work is to explore and demonstrate the ability of a Convolutional Neural Network (CNN) to classify cells pictured via brightfield microscopy without the need of any feature extraction, using a minimum of images, improving work-flows that involve cancer cell identification. Design/methodology/approach: The methodology involved a quantitative measure of the performance of a Convolutional Neural Network in distinguishing between two cancer lines. In their approach, they trained, validated and tested their 6-layer CNN on 1,241 images of MDA-MB-468 and MCF7 breast cancer cell line in an end-to-end fashion, allowing the system to distinguish between the two different cancer cell types. Findings: They obtained a 99% accuracy, providing a foundation for more comprehensive systems. Originality/value: Value can be found in that systems based on this design can be used to assist cell identification in a variety of contexts, whereas a practical implication can be found that these systems can be deployed to assist biomedical workflows quickly and at low cost. In conclusion, this system demonstrates the potentials of end-to-end learning systems for faster and more accurate automated cell analysis.