The first important thing Rich Gliklich did with his career was interrupt it. During medical school, with the tidy route from student to physician already laid out, he stepped away for a year to investigate a bothersome question: how could anyone know whether medical care had actually produced a good result? A clinical trial could tell you what happened inside a carefully bounded experiment. Ordinary care was wider, noisier and badly organized. It was also where nearly everyone lived.
That year became the hidden spine of everything that followed. Gliklich returned, earned his medical degree at Harvard, trained in otolaryngology and facial plastic and reconstructive surgery, and joined academic medicine. Yet outcomes remained the itch. At Massachusetts Eye and Ear, he developed a clinical outcomes research unit and worked on the practical mechanics of asking large groups of patients what happened after treatment.
The vocabulary sounds commonplace now. Patient journeys, real-world data and evidence generation have become conference-hall dialect. In the 1990s, the infrastructure was thinner and the pitch less fashionable. Medicine had plenty of records, but records are not the same thing as evidence. The first are accumulated. The second must be designed.
A laboratory with a business plan
Gliklich has described himself at the time as a happy academic. Then a business-development director suggested that his research operation might work as a company. He knew medicine, research and some technology. He has been frank about the missing part: he knew almost nothing about business.
Outcome Sciences emerged from that imbalance in 1998. It was less a leap from academia than an attempt to give academic methods operating machinery. The company designed registries and observational studies, the unromantic plumbing required to follow patients over time, compare treatments and watch for patterns that the original trial had been too small or too short to reveal.
The business grew for roughly twelve years and came to serve more than 2,500 healthcare organizations, as well as many of the largest life-sciences companies. Its timing was instructive. When the limits of tightly controlled studies became impossible to ignore, registries moved from a methodological side street toward the main road. Gliklich was already there, armed with the rare competitive advantage of having worried about a dull problem before it became urgent.
He also helped turn practice into doctrine. Beginning in 2007, he served as an editor of Registries for Evaluating Patient Outcomes: A User's Guide, commissioned by the Agency for Healthcare Research and Quality. By the fourth edition, the guide had become a durable reference for people designing, running or judging registries. Its subject is quality, which is another word for all the ways a persuasive number might still be wrong.
A related federal effort placed him in the middle of a harder argument: deciding which outcomes should count across different conditions. The Outcomes Measures Framework brought together agencies, clinicians, payers and patient organizations around broad domains such as survival, clinical response, events, patient-reported outcomes and resource use. The list looks almost obvious after someone has negotiated it. Before agreement, every specialty can arrive with its own ruler.
Standardization carries a faint whiff of the stationery cupboard, but it changes what can be learned. If two studies define improvement differently, pooling their findings may produce arithmetic without meaning. If hospitals record the same result in incompatible ways, benchmarking rewards confidence more readily than performance. Gliklich's work kept returning to this precondition: comparison becomes useful only after the things being compared have been described with care.
One question, three professional lives
The successful exit that felt unfinished
Quintiles acquired Outcome Sciences in October 2011. Gliklich became a leader of its late-phase business and remained through the parent company's 2013 public offering. On paper, the arc had reached its ceremonial ending: laboratory, startup, scale, acquisition, public company. Gliklich's own telling is more useful because it leaves room for a less photogenic emotion.
He felt remorse. The checking account was larger, he later recalled, but the sale did not feel like arrival. He knew he would found again. It is a small admission with a large explanatory radius. Some people enjoy creating an institution more than possessing the proof that they created it.
There was an intermission. He spent time at General Catalyst as an XIR, a role that let him examine healthcare companies from the investor's side of the table. The view sharpened a new possibility. Electronic health records were proliferating, computing had improved, and machine learning offered ways to organize information that once required armies of human abstractors. The old outcomes question could be asked at a different scale.
The second act begins with old records
OM1 began in 2015. The name compresses a manifesto into three characters: outcomes, measurement, first. The company assembles longitudinal clinical data, builds specialty networks and applies analytical models to research and decision-making. In 2021, it announced an $85 million financing to expand those data networks, its registry work and its AI platform.
The specialty networks are a practical answer to the shallowness of scale. A vast dataset can still be clinically thin. By working with specialists in fields such as rheumatology, dermatology, cardiology and otolaryngology, OM1 sought denser records: the measures, symptoms and treatment details that make a disease legible. Bigger is useful. Deeper is what allows the interesting question.
Gliklich's pitch has stayed notably consistent beneath the changing technology. A trial asks participants to generate fresh information according to a protocol. Routine care already generates enormous amounts of information, from claims and laboratory results to narrative notes. If more of that existing material can be standardized and extracted automatically, researchers can reserve expensive active collection for facts that truly are missing.
The distinction between passive and active collection sounds bureaucratic until someone pays the bill. Manual chart review is slow, expensive and difficult to scale. Gliklich argues that automation can reduce the burden on study sites while enlarging the population a study can observe. More people, followed for longer, can make rare events visible and show how outcomes vary beyond the polished boundaries of a trial.
A phenotype made from the trail
OM1's digital-phenotyping work looks across many dimensions of longitudinal data to identify patterns, possible subtypes and likely trajectories. Gliklich describes it as a route from the record of past care toward a more personal next decision.
The ambitious piece is prediction. OM1 calls its approach digital phenotyping: looking across a long clinical trail for a pattern that can characterize a patient more precisely than a broad diagnosis alone. Gliklich has compared the output to a fingerprint and the concept to a data-driven cousin of genotyping. A model might identify a subtype, estimate a likely trajectory or indicate which patients resemble those who responded to a treatment.
This is where the old registry builder remains visible inside the AI founder. Prediction invites theater; outcomes research insists on definitions. What exactly is the endpoint? Which population produced the model? What information was absent? Does a comparison account for how ill people were at the beginning? The clever model arrives last. First comes the stubborn work of making the evidence trustworthy.
A zigzag with a fixed point
Gliklich has offered an appealingly ornithological warning about careers. Many people, he says, behave like ducklings following the larger bird in front. His own path supplies the alternative: step away from medical school, return to surgery, turn a research unit into a business, sell it, try venture capital, then found again. The route zigged and zagged. The question held still.
His other metaphor comes from comics. Asked which superpower best resembles data science, he chose precognition, the ability to see a likely future. It is a charming answer from a man whose professional life has been devoted to the limits of prediction. Waverider, the comic-book character Gliklich cited, can monitor streams of time. Healthcare gets scattered records, missing fields and doctors who quite reasonably prefer caring for people to formatting datasets.
That gap tempers the futurism. Gliklich's public aspirations are large: use real-world evidence to personalize care, help locate people whose conditions have escaped notice, and show a clinician what happened to comparable patients after a particular choice. Yet the means remain procedural. Measure. Standardize. Compare. Explain.
He still holds an academic appointment at Harvard Medical School and directs a clinical outcomes research unit at Massachusetts Eye and Ear. The parallel roles matter. They keep his entrepreneurial story attached to the place where the question began: the encounter after which somebody will want to know whether the intervention helped.
In 2026, the circuit curved back to registries. At the DIA Global Annual Meeting, Gliklich joined a panel on the difference between a registry that produces decision-ready evidence and one that merely accumulates data. The title was cheeky, “basket trial or basket case,” but the concern could have come from the first year of his career. Technology changes the size of the basket. It does not absolve anyone from inspecting what went into it.
The glamour in healthcare technology usually gathers around invention. Gliklich's career makes a case for follow-up. The operation ends. The prescription is filled. The trial closes. Then life resumes, disorderly and largely unobserved. He has spent three decades building ways to look again.