Intelligent health monitoring: Why community data must be part of the system

Posted on

by

ITPC at IAS2025

IAS 2025 Advancing HIV Monitoring: Community Leadership, Big Data Science and Systems Integration Symposium Presentation

Solange Baptiste, ITPC Executive Director

It is imperative to think about the promise, the pitfalls and the problems with AI as we think about our context, about data and monitoring. AI will help to process massive volumes of health data faster than any human expert, identifying patterns, predicting outbreaks, optimizing supply chains.

We are beginning to understand AI’s potential for health improvements, from vaccine research and development to personalizing care and increasing efficiency – all of which is important with limited resources.

The pitfall is that AI is only as good as the data it’s trained on. Biases in data will lead to biases in care – trash in equals trash out. Most community realities are underrepresented or missing entirely in national data sets and without these contexts, the predictions risk being inaccurate and also unjust.

The problem here is something we’re calling data poverty. There is structural under investment in community data systems, resulting in many populations being invisible to AI. No data does not mean no need – it means no one was paid to listen to those people. If we don’t include the lived experience, AI will reproduce the blind spots.

What we have is something nice and fancy and shiny and new that’s reproducing the inequities we already have, only it’s doing it much faster. Without community leadership, AI risks becoming just another top-down system that is faster at replicating the inequities that persist.

But what is the goal of big data? I would suggest that it is intelligence: the ability to turn fast, diverse data into timely, actionable insight that improves decisions, outcomes and equity and that means seeing the unseen, responding early and allocating wisely. The real value of big data is not having more numbers, the value lies in knowing what those numbers mean, when those numbers matter, and who those numbers serve.

Data without people isn’t intelligence, it’s noise. What real intelligence requires is scale plus speed, but also context.

We already have programmatic and strategic intelligence – those things we always know, like big data, HMIS – which gives us scale and breadth, and without it you miss the full picture. It is compromised, increasingly fragmented and politicized.

We are witnessing the exponential rise of artificial intelligence which gives us speed and pattern detection, and without it we risk being slow to respond.

Finally, there is community intelligence which gives us context and grounded prioritization without community inetelligence, you act on the wrong assumptions.  Community intelligence is completely under-invested and often invisible.

Put another way, the intelligence layer that the system forgets is the community data or CLM. CLM fills in reporting blind spots. It supports better value for money decisions. It tracks last mile failures and informs reallocation. It provides the why and the what and the alignment with community needs.

Let’s not waste the current crisis. The goal is not just to centralize data; it is to integrate perspectives. Monitoring systems must evolve to match the reality of integrated service delivery. AI and DHIS2 are not enough without community-generated signals. CLM is not a separate stream; it’s a grounding layer for real-time course correction (early warning) and systems feedback. If communities don’t live in silos, our data systems shouldn’t either.

Health systems without community data are ineffective and incomplete and we have the power to fix this. The question is, will we?