The Preventive Care Gap: Why We Wait to Spend Until It’s Too Late

I’ve been thinking more about my health lately. I’m in my 40s, and like many women my age, I’ve started asking new questions: How will my health change as I age? How much should I save for care?

That curiosity led me down a data rabbit hole — and what I found honestly shocked me.

It turns out that the U.S. spends a massive amount of money on people’s healthcare — but mostly after they turn 65. We spend more than $1 trillion every year into Medicare, which is the federal program that covers health insurance for older adults.

(Source: https://www.reuters.com/markets/us/us-budget-deficit-tops-18-trillion-fiscal-2024-third-largest-record-2024-10-18/)

But when I looked at how much we invest in preventing illness in younger people, the difference was shocking.

CDC spent on chronic disease prevention that same year was $1.4 billion

(Source: https://www.tfah.org/report-details/funding-2024/)

From a resource allocation perspective, the variance is staggering. We’re talking about a 700x difference in spend — yet the ROI on prevention is higher. If this were a portfolio, no investor would structure it this way.

It kind of blew my mind. Because we know that taking care of our health earlier in life can stop so many problems later.

What if we matched just a fraction of what we spend on care after age 65 with investments in keeping people healthier before they even need it?

Imagine more access to nutrition coaches, mental health therapists, or fitness programs for people in their 20s, 30s, and 40s. Imagine if we normalized early screenings and made them free and accessible. We could catch things early, reduce suffering, and save money in the long run. 

As someone who works with data and models every day, I couldn't help but approach these questions like I would any system — looking for patterns, gaps, and missed signals.

Here are three ideas we could turn into reality, starting now.

 1. Use SDOH data to Find Risk Before It Becomes Reality

We already know that Social Determinants of Health (SDOH) — like where someone lives, their income level, education, transportation access, and more — affect 30–55% of health outcomes.

So… what are we doing with that knowledge?

Right now, not nearly enough. But we could.

Imagine this: we build patient cohorts and lookalike models using SDOH data to identify people who are most likely to develop a certain condition before it happens.

This is where predictive modeling shines. By training lookalike models on patients with known diagnoses and layering in environmental and socioeconomic data, we can flag high-risk individuals with precision — even before a claim is ever filed. Like communities living near industrial zones or mining sites — they may be more prone to certain respiratory or oncological conditions. 

What if we proactively deployed preventive screenings, mobile clinics, or health education programs in these areas?

Instead of waiting for a diagnosis at Stage 4, we catch it at Stage 0. That’s not just better medicine — it’s smarter economics.

2. Use Foot Traffic data to Understand Who’s Actually Getting Care

Here’s where it gets even more interesting.

Large healthcare systems like Integrated Delivery Networks (IDNs) are central to how care is delivered. With anonymized foot traffic data, we could uncover the patterns.

By blending foot traffic data from digital geolocation with IDN affiliation metadata, we can build inferred care maps — essentially reverse-engineering healthcare journeys in real time. Then, we apply clustering or anomaly detection to surface where behavior doesn't match expected access patterns.

Say we notice a rural infusion center has high foot traffic — but local SDOH data shows low education levels and limited support systems. That mismatch could point to gaps in health literacy, missed follow-ups, or unaffordable prescriptions post-treatment.

That’s a chance to intervene. Maybe with a community navigator program or financial assistance tools or just clearer communication.

3. Stop Thinking About Drugs in Isolation. Think in Systems.

Here’s a radical shift: instead of analyzing access at the pharma brand level, what if we looked at it from the treatment center up?

Because these centers aren’t just where patients go to receive treatment. They’re also where the healthcare system comes together: providers, payers, care managers, caregivers, and patients all meet in this one physical space.

From a systems perspective, treatment centers are node points — places where multiple data streams converge: SDOH, claims, clinical data, logistics, and human interaction. We need to analyze this like a network — not a list. Graph modeling, for example, could reveal access deserts or patient drop-off points we currently miss.

That gives pharma, health systems, and public health agencies a shared map. And when everyone shares the same map, collaboration becomes possible.

It’s about reimagining what healthcare could look like — not someday, but now.

We could build a system that doesn't just treat disease, but prevents it. Predicts it. Prepares for it.

That’s the kind of healthcare I want for my future — and yours too.

As someone working at the intersection of data strategy, AI and public health, I see so much potential in applying machine learning not just to optimize treatment — but to rethink who gets care, when, and why. This is the future of health equity — and it’s entirely within reach.

#HealthcareStrategy #HealthEquity #DataForGood #PredictiveAnalytics #PublicHealth

#HealthTech #PreventiveCare #MachineLearning #DataScience

Note:The views expressed here are my own and do not reflect those of my employer.