At the Centre for Evaluation, we are considering the role of Artificial Intelligence (AI) in the evaluation of complex public health interventions and the implications of using AI in evaluation, on our efforts to decolonise evaluation and enhance participatory practices in evaluation.  

Artificial intelligence is increasingly being used in the health sector to improve the delivery of health services, strengthen health systems, inform health policy and decision making. Normative organisations and national governments are considering the ethical implications of using AI in the health sector and tackling issues of equity, power and autonomy.  

However, there has been limited attention paid to the role of AI in process and impact evaluations of complex public health interventions to date. To address this, we convened an expert workshop in December 2024.

We reviewed the evidence of using AI in the evaluation of complex public health interventions and to explore the transferability of AI utilisation from other health sectors to evaluation designs. A full agenda of this workshop is available below.

Agenda for workshop: Harnessing Artificial Intelligence in the evaluation of complex public health interventions

Plenary Welcome and introductions LSHTM’s AI strategy Overview of AI: what it is and is not and key considerations Ethics and governance of artificial intelligence for health

Expert Panel: AI in evidence synthesis Presentation 1: Developing guidance for the responsible use of AI in evidence synthesis Presentation 2: Application of machine learning for evidence synthesis in climate change and health: screening and labelling Presentation 3: Using AI to automate descriptive mapping for evidence synthesis Presentation 4: Developing machine learning systems for extraction and prediction based on an ontology of behaviour change interventions Facilitated Q&A

Interactive session 1: Open Mic

Expert Panel: AI in process evaluation Presentation 1: Using AI in evaluations: benefits and challenges Presentation 2: AI approaches to using visual data in field evaluations Presentation 3: Machine learning and natural language processing to develop programme theory of change and theory-driven evaluation Presentation 4: Scaling behavioural science with AI: Rapid qualitative interviews with UNICEF Türkiye Facilitated Q&A

Interactive session 2: Opportunities & Challenges

Alongside this workshop, AI is an important, and recurring, topic of focus for the Centre regular series of seminars. Find out more about our specific seminars on the topic, and access the recordings, below.

Dr Charis Wong: AI-enabled systematic review platforms: how well do they perform?

Seminar hosted February 2026 at LSHTM which discussed the preliminary findings from research - led by Dr Charis Wong - that evaluated the performance of three platforms (Elicit.AI, Nested Knowledge and Scispace) across different systematic review stages.

Watch recording.

Professor James Thomas: Recommendations and guidance on responsible AI in evidence synthesis

Seminar hosted in 2025 at LSHTM which explored efforts among evaluation experts to create new guidance on the responsible use of AI in evidence synthesis research. The speakers also discussed ways to evaluate the use of AI tools and understand the capabilities for evidence synthesis tasks.

Watch recording.

Professor James Thomas: New 'AI' technologies for evidence synthesis: how do they work, and can we trust them?

In 2024, the Centre hosted a seminar to address the explosion of interest in generative large language models (LLMs) from the past year.

Professor James Thomas based at the EPPI Centre, UCL gave a basic lesson about LLMs and what they were capable of (as of January 2024) and how they could be useful for research as well as critically assessing their (sometimes outlandish) claims of their utility.

Watch recording.

2024 panel discussion on AI in impact evaluation

During the panel discussion, colleagues discussed the range of extensive use of AI in statistical analysis and opportunities to share good practice with evaluators including:

• the use of AI to design interventions, enhance recruitment and retention, to maximise sample size for example with digital twins in the control groups
• the use of computational methods to measure the impact of expenditure on development outcomes
• the use of AI for genomic surveillance
• the value of machine learning but the importance of understanding the ‘black box’ of machine learning tools

We also heard about ways in which AI is being used not only to transcribe, translate and code qualitative data, but also to collect and analyse it. We heard about the value of AI in processing images for classification, object detection, and scanning, processing and summarising quantitative data. These discussions highlighted the potential of adopting or adapting the existing AI methods for evaluations to support:

• measuring implementation
• mechanisms of impact and context factors

During the workshop we highlighted opportunities and challenges that the use of AI provides in process and impact evaluation. We are in the process of drafting a commentary summarising the highlights of the workshop and are making plans to progress this work over the coming years. If you are using AI tools in the evaluation of complex public health interventions, please get in touch with us and share your experiences. Please write “AI in public health evaluations” in the title of your email to [email protected].

We look forward to continuing this discussion with experts in public health, evaluation and AI and beyond.