100 Health Needs a Year. Three Doctor Visits.

How many times in a year do you think you have a health need?

Before you answer, consider an ordinary week. A headache on a Tuesday afternoon. A new pain in your shoulder that was not there last month. An upset stomach. A night of poor sleep, or three. A stretch of anxiety about work, money, or family. Few of these would prompt a call to a doctor's office, but each one is a health need: a moment when your body or mind asks for attention and you have to decide what to do about it.

Most people, asked to estimate, offer a number in the single digits or low teens. Six decades of research suggest the real number is closer to 100.

What the research shows

This question has been studied since the 1960s under the name "the symptom iceberg," and the findings are remarkably consistent. A UK population survey found adults experience an average of 3.66 symptoms every two weeks, roughly 95 episodes per year. A Danish study of nearly 50,000 adults found that 91.5% reported at least one symptom in the past month. The classic Ecology of Medical Care model found that in any given month, 800 of every 1,000 people are symptomatic. An estimate of roughly 100 health-need events per adult per year is well supported by the evidence, and probably conservative.

The more important question is what happens to those events.

Every symptom is a decision

Each of those 100 events ends the same way: with a decision, usually made quickly and without much reflection. Ignore it and see if it passes. Manage it yourself. Or seek help.

The research shows how those decisions resolve, and the result is a pyramid. Of 100 annual events, approximately 50 are ignored or resolved through watchful waiting. Another 38 are self-managed with over-the-counter medicine, home remedies, apps, and information from the internet. Roughly 8 receive professional input without a visit: a portal message, a nurse line, a pharmacist consultation, or an e-visit. Only 3 to 4 become physician office visits, a figure that aligns almost exactly with the federal benchmark of 3.2 visits per person per year.

The implication deserves emphasis. Ninety-six of every 100 health needs resolve without ever reaching a physician. The formal healthcare system, which the country spends more than $4 trillion a year operating, engages with only a small fraction of the health decisions people actually make.

The advice layer has a new owner

For most of the internet era, the self-managed layer belonged to Google and WebMD. Over the past 18 months, that has changed at a pace without precedent in the history of health information.

OpenAI reports that more than 40 million people ask ChatGPT health questions every day, and that one in four weekly users sends at least one health prompt each week. Microsoft reports 50 million health queries per day on Copilot, where health is the most common topic among mobile users. Google has embedded AI Overviews into its search results, where they now appear on the majority of health searches and 93% of symptom searches; when an AI Overview appears, most users read the AI-generated answer and click no further. Taken together, well over one billion health questions per day are now answered first by generative AI.

Surveys suggest a smaller footprint, with roughly a third of adults reporting they used AI for health information in the past year. Those surveys are accurate as far as they go, but they miss a structural change: a person who searches Google and reads the AI-generated answer at the top of the page does not report having used AI. The categories of "searching online" and "using AI" have effectively collapsed into one.

The per-capita arithmetic is significant. US chatbot health interactions alone now run on the order of 110 to 150 per adult per year, meeting or exceeding the entire annual base of 100 health-need events. On average, every American health need now touches an AI system at least once. The advice is also consequential: 42% of adults who use AI for health advice do not follow up with a professional, and an estimated 14 million US adults per month report skipping a provider visit based on AI guidance.

Self-management now comes with professional-grade data

A second expansion of the self-management layer is underway, and it involves data rather than advice. Testing that once required a physician's order is now available directly to consumers. Function Health, which offers a membership covering more than 160 lab tests for $365 a year, has completed more than 50 million lab tests since 2023, serves hundreds of thousands of members, and raised $298 million in November 2025 at a $2.5 billion valuation. Quest and Labcorp now sell panels directly to consumers, and Hims & Hers, Oura, and Whoop have each added consumer lab testing through partnerships with Quest. Demand for full-body imaging follows the same pattern, with one provider reporting a waitlist of more than 100,000 people.

Wearables have made the monitoring side just as accessible. According to Rock Health's 2025 consumer survey of 8,000 US adults, 57% now own at least one wearable or connected health device, and 46% own a wearable specifically, up from 13% a decade ago. Continuous glucose monitors are now sold over the counter, including on Amazon. Heart rhythm, sleep staging, blood pressure, glucose, and recovery scores stream continuously from devices people wear all day.

Each of these data streams feeds the same decision queue. A biomarker flagged as out of range, a sleep score that dips for a week, a glucose curve that spikes after lunch: every reading is a new health-need event that must be ignored, self-managed, or escalated. And the interpretation increasingly happens inside the same AI advice stack, where lab results and wearable exports are uploaded for explanation. The practical effect is that the self-management layer is no longer just where minor symptoms go to resolve. It is acquiring the tools of a clinical workup, generating signals at a volume no annual physical can match, and expanding the base of the pyramid in the process.

Your feed is adding to the pile

There is one more force at work, and it does not wait for you to notice a symptom. Search engines and AI chatbots respond to needs you already have. Social media generates new ones. The video suggesting your hormones may be out of balance. The post implying that your afternoon fatigue is not normal. The sleep content that turns an ordinary restless night into a deficiency worth investigating. Algorithmic feeds surface health content users never sought, and incidental exposure makes people 4 to 6 times more likely to begin seeking health information themselves.

The effect is measurable and not always benign. Among Gen Z, one in three cites TikTok as their primary source of health information, despite research showing that fewer than half the claims in popular #ADHD videos are clinically accurate. Ordinary experiences such as distractibility and fatigue are increasingly being reframed as symptoms at scale. Every piece of pushed content adds another item to the queue of decisions each of us is already managing: ignore it, self-manage it, or seek help. The 100-event annual base should be treated as growing, not static.

What this means for a health system

Three observations follow from this research.

First, the triage moment has moved. The moment when a person first interprets a symptom and chooses among those three options now belongs predominantly to AI. We wrote earlier this year about the front door to healthcare moving. This research quantifies the shift: AI systems now process in roughly two weeks the interaction volume the entire US office-based physician system handles in a year.

Second, the highest-leverage layer receives the least investment. The 8 events per person per year that reach nurses, pharmacists, and portal inboxes represent more than double the volume that becomes physician visits, and they are the fastest-growing part of the formal system. Portal messages rose 157% in a single year during the pandemic and have not returned to prior levels. Many systems now use AI to draft portal replies, which means generative AI already operates on both sides of the professional boundary. Making this layer fast, trustworthy, and genuinely useful is the most direct way a health system can compete for the triage moment.

Third, the 96 events that never reach a physician are where the patient relationship is built or lost. Each one represents a decision, often made in seconds, about whether the health system is relevant to that person's life. Today, in most of those moments, the health system is absent.

At Ballad Health, serving a rural region where access is constrained and trust matters deeply, we view these 96 moments as the defining competition of the next decade. Our intention is to be present for them.