PhD student publishes systematic review in prestigious journal

IMPALA PhD student William Nkhono has just authored the first systematic review (to our knowledge) of machine learning models for hospitalised children in low- and middle-income countries. The article appears in the high-impact publication eClinicalMedicine, a Lancet Discovery Science journal.

“The review provides a baseline for projects such as IMPALA by clarifying where prior efforts fell short and where contextual novelty can be introduced,” says William.

“Despite the growing enthusiasm for AI in global health, the literature is fragmented and skewed toward high-income settings. I wanted to map what has been done, identify use cases and expose gaps in clinical relevance, data quality and implementation readiness of AI.”

To help integrate the tools into clinical practice, William and his co-authors developed 8 recommendations for developing machine learning applications in low- and middle-income countries:

1. Think and design ahead

Map barriers and facilitators early in the machine learning development process for clinical use. Apply them to draft a development plan.

2. Plan for data availability and quality

Address data and technological aspects immediately, such as investing in digital data collection and server capacity to improve data quality, and developing datasets for model development.

3. Keep the clinical application and end user in mind

Invest early in characterising user needs, acceptability, workflow integration, and economic sustainability.

4. Focus on implementation

Prioritise addressing major health challenges in low- and middle-income countries as they will have the greatest impact. Report all necessary performance metrics, including those that are clinically relevant.

5. Keep it simple and explainable

Match machine learning methods to clinical needs and available resources,
Use simple, interpretable models when explainability is required, as this would increase clinical uptake,
Invest in making outputs explainable and actionable to end users.

6. Leapfrog while learning

Anticipate ethical and regulatory challenges as machine learning approaches clinical use,
Design machine learning systems with compliance in mind and contribute proactively to best practices and regulations,
Build on the experience of others, including those in high-income countries.

7. Increase visibility and equitable collaboration

Establish equitable international networks to develop, validate, and share models and resources,
Consider how your current research can contribute to lasting data infrastructures and local and regional AI strategies.

8. Empower policy and governance actors

Policymakers, regulatory bodies, ethics committees, funding agencies, and international organisations such as the WHO should prepare a facilitating environment for developing and integrating machine learning into healthcare in a safe, responsible, and effective way.

Read the full paper online (open access): “Machine learning for predicting clinical outcomes of hospitalised children: a systematic review of applications in low- and middle-income countries”.