Your Doctor's New Colleague Doesn't Have a Pulse

According to Stanford's Institute for Human-Centered AI, researchers have built a model that can predict future disease risk from a single night of sleep.

One night. Your breathing patterns, heart rate variability, movement during REM — fed into a model that spots patterns invisible to human eyes. And apparently those patterns say things about what's coming for your body that no blood test currently can.

If that doesn't make you pause, consider this: by the end of 2026, almost 90% of American hospitals will have adopted AI-driven diagnostics, according to a DashTech healthcare analysis. (The picture elsewhere is starkly different — the WHO estimates that fewer than 30% of hospitals in sub-Saharan Africa have reliable internet access, let alone AI infrastructure.) In the United States, AI-generated clinical notes are now being accepted by CMS and major insurers for billing purposes, according to Mass General Brigham researchers. And according to Notable Health, AI-powered EHR integration in US hospital systems could handle approximately 50% of routine administrative work, saving the average American physician 15-20 hours per week. Britain's NHS is running its own AI pilots, though adoption has been slower due to procurement rules. India's National Health Authority is exploring AI for its Ayushman Bharat scheme, which covers 500 million people — a scale that would dwarf any US deployment.

The stethoscope has competition.

Luke Would Have Questions

Luke was a physician. We tend to forget that about the author of the third Gospel and Acts — he was a doctor first, a writer second. His Gospel has more medical detail than the others. He notices healings differently. He's clinically precise.

I sometimes wonder what Luke would make of an AI that reads polysomnography data and predicts cardiovascular risk five years out. I think he'd be fascinated. I also think he'd ask the question we should all be asking:

Who benefits?

Because technology never distributes itself evenly. The first stethoscopes served wealthy patients in European cities. MRI machines went to research hospitals decades before they reached rural clinics. And right now, AI diagnostic tools are being deployed primarily in well-funded hospital systems in developed nations.

The gap between Stanford Medicine and a clinic in rural Mississippi is real. The gap between Stanford Medicine and a clinic in rural Malawi is a chasm. Between Stanford and a community health worker in Bangladesh navigating flooded roads with a phone? That's a different universe entirely. The technology is global; the access is not.

'Heal the Sick' — With an Algorithm?

Jesus told his disciples to "heal the sick" (Matthew 10:8). It was part of the commission. Not optional. Not aspirational. A command.

For centuries, the church has been one of the primary drivers of healthcare worldwide. Christian missionaries built hospitals across Africa, Asia, and Latin America. Religious orders staffed clinics. The Salvation Army, Catholic Charities, and countless local churches have been healthcare providers for communities that nobody else served.

So when an AI can detect disease earlier, diagnose more accurately, and free physicians to actually be present with patients instead of buried in paperwork — isn't that aligned with the mandate?

I think it is. With caveats the size of a radiology department.

The Caveats

First: AI diagnostics are only as good as their training data. If the data skews toward certain populations (which it does), the AI will be better at diagnosing some groups than others. A widely cited 2021 study in The Lancet Digital Health found that dermatology AI systems performed significantly worse on darker skin tones — largely because training datasets were drawn predominantly from light-skinned patient populations in North American and European hospitals. Five years later, Chief Healthcare Executive reports that the bias gap has narrowed but not closed.

This isn't a technical problem. It's a justice problem. And the church — which claims to serve a God who "shows no partiality" (Romans 2:11) — should be among the loudest voices demanding equity in AI healthcare.

Second: efficiency is not the same as care. If AI saves a physician 15 hours a week, the question is what happens with those 15 hours. Do they see more patients? (Good.) Do they spend more time with each patient? (Better.) Does the hospital system reduce staff because "AI handles it now"? (Concerning.)

The parable of the Good Samaritan isn't a story about efficient wound assessment. It's about someone who stopped. Who got close. Who touched. AI can't do that. If it frees humans to do more of it, praise God. If it becomes an excuse to do less of it, we have a problem.

Third: sleep data is intimate data. Your breathing patterns, your heart rate at 3 AM, your restlessness — this is vulnerable information. And the question of who owns it, who profits from it, and who can access it matters enormously.

We wouldn't be comfortable with a stranger watching us sleep. We should think carefully about what it means to let a corporation's algorithm do it.

Common Grace in a Neural Network

Reformed theology has a concept called common grace — the idea that God distributes gifts, knowledge, and creativity broadly across humanity, not just to believers. The surgeon who saves your life may or may not share your faith. The grace is in the gift, regardless of the giver's theology.

AI in medicine feels like common grace to me. It's knowledge accumulated by thousands of researchers, encoded into systems that can spot a tumor shadow that a tired radiologist might miss at 2 AM. That's a gift. An imperfect, complicated, commercially-motivated gift — but a gift.

The question isn't whether to accept it. It's whether to steward it.

What the Church Can Actually Do

This isn't just a "pray about it" situation. (Although pray about it.) Churches with healthcare ministries should be asking:

• Are the AI tools we use tested across diverse populations?
• Are we advocating for AI healthcare access in underserved communities?
• Are we helping our congregations understand what these tools can and can't do?
• Are we preserving the human element — presence, touch, prayer — that no algorithm can replicate?

Because the future of medicine is almost certainly AI-augmented. The question for the church is whether we'll be at the table shaping how that augmentation works, or whether we'll show up late and wonder why nobody asked us.

(They're not going to ask us. We need to show up.)

The Doctor Who Wrote a Gospel

Luke didn't stop being a physician when he became an evangelist. He brought his medical eye to the story of Jesus. He noticed the hemorrhaging woman. He recorded the healing of the ten lepers with clinical specificity. He understood that bodies matter because the God he served took on a body.

If Luke were practicing today, I suspect he'd be fascinated by AI diagnostics. I also suspect he'd still sit at the bedside. Because the algorithm can read the scan, but it takes a person to hold the hand of someone who just got the results.

That part is still ours.