New Physician Careers in the World of AI and Digital Health

A guest essay by Marshall Frieden

During residency, I saw how much frustration came from the technology surrounding patient care. Electronic health records were hard to work with, and when systems could not share information in a usable way, those interoperability gaps added to the burden on physicians and patients. I wanted to improve that experience while continuing to practice medicine.

I think of it as continuing to fly the plane while helping make it better for the other pilots and the passengers. Test pilots play that role in aviation; clinical informaticists play it in digital health. For me, that meant caring for my own patients while improving the records and tools my colleagues relied on to care for theirs. Steady, iterative improvement to the EHR and the platforms around it could remove some of the friction I saw in residency and give physicians more time to be present with patients.

During my clinical informatics fellowship, I started working out how to turn that goal into a career. I needed to decide how much clinical practice I wanted to keep, what kind of informatics work interested me, and how that work would be supported in an actual appointment. Sorting the options into academics, health system operations, and industry gave me three useful starting points. The paths overlap, but each offers a different way to combine clinical experience with work that reaches patients beyond the ones you see face to face.

Academics

Academic informatics suits physicians who want to study a problem, teach others to solve it, and stay involved in patient care. The key question is how each activity is paid for. Research funding, an institutional informatics appointment, and an educational leadership role are different ways to buy time, and each comes with different expectations.

A quick note on full time equivalent, or FTE: a 0.2 FTE informatics assignment generally means 20% effort. Always ask how that becomes protected time on your actual schedule.

Build research into the appointment

If you are drawn to research, start with a clinical question and a team that can help you study it. Does a decision support tool change care? Does an AI model perform equally well across patient groups? How does a new workflow affect clinicians? Learn the methods your question requires, and find collaborators for study design, data access, and analysis.

Grant funding can cover part of your effort, but “getting a grant” does not mean your full salary is covered or that your clinical schedule changes. NIH career development awards illustrate the gap: most K awards require at least 75% professional effort, while NIH funds only part of the salary. Before accepting a research commitment, ask who covers the rest, which clinical duties will be reduced, and what happens when the award ends.

Early on, you may join an established investigator’s project rather than lead your own. As your role grows, track the contributions you can point to: a useful dataset, a sound evaluation, a publication, an intervention that changed practice. Ask how that work counts toward promotion at your institution rather than assuming every valuable operational project meets research expectations.

Secure institutional time for informatics and teaching

An academic health system may also fund your work directly as a physician informaticist or clinical leader, supporting implementation, decision support, quality improvement, or oversight of new technology. That work can look a lot like the operations path below even when the appointment is academic. Ask who pays for the effort, who sets priorities, and whether scholarship is expected on top of implementation.

Education is another real route. You might design an informatics curriculum for students or residents, supervise projects, teach fellows, or eventually lead a fellowship. A curriculum proposal is stronger when it names the learners, the teaching time, the assessment, and the person responsible for maintaining the course.

Teaching time must be negotiated, not assumed. For accredited clinical informatics fellowships, the ACGME requires dedicated program director support starting at 0.20 FTE for smaller programs and rising with program size. That support often comes as release from other duties, not additional salary. An occasional lecture and ownership of a longitudinal curriculum are very different commitments, so match the time request to the responsibility.

Grow toward research and AI leadership

For physicians who enjoy building the infrastructure behind research, a chief research informatics officer (CRIO) role is one senior destination. The remit is broader than running a study: it means enabling other investigators, supporting clinical trials and translational science, and aligning research needs with enterprise technology.

The chief health AI officer is a newer leadership role that a growing number of academic health systems have created, several of them filled by physicians. The work spans AI strategy, governance, evaluation, and implementation. It reflects a real need for leaders who can judge whether a tool belongs in clinical practice and how it should be monitored once it is there.

Neither role is an automatic next step after fellowship or a grant, and neither is reserved for physicians. Titles and reporting lines vary widely. If this work interests you, look for experience leading across clinical, research, and technical teams, with real accountability for decisions and their consequences.

Health system operations

Operations fits physicians who enjoy taking a problem clinicians hit every day and fixing the system underneath it. A good starting point is an informatics, medication safety, quality, documentation, or decision support committee. Pick one where you can contribute to a defined problem rather than collecting memberships.

Bring a concrete proposal. A team may need a dashboard to understand delayed follow up, a better referral workflow, or an order set that reflects an agreed clinical pathway. In an Epic environment, that could mean improving SmartSets or other content with the analysts and clinicians who own the build. The work starts with understanding the workflow and agreeing on a measure of success. Changing the template is only part of it.

For analytics, ask what the data actually represent, whether the denominator is right, and who will act on the result. For EHR content, plan for clinical review, testing, implementation, and maintenance. For any proposed alert, ask whether it helps the clinician in that moment or just adds another interruption. The goal is an improvement patients and care teams can feel, with evidence that it worked.

Turn projects into a defined role

A successful project gives you something specific to bring to a clinical or informatics leader: the problem, your contribution, the result, and what still needs an owner. That is the basis for asking about a physician informaticist appointment, a service line role, or protected project time. Volunteering helps you learn the work, but growing operational responsibility needs an agreed scope, support, and accountability. Otherwise it becomes a second job done after your clinical one.

The next step may be an assistant or associate chief medical information officer (CMIO) role, though organizations use those titles differently. Some divide informatics leadership by ambulatory care, inpatient care, specialty, or population health, and many pair the role with a substantial clinical commitment, often around half time. Treat any specific split as one job design, not a standard.

Your clinical background can point you toward a domain, but the job also requires learning how other people work. An ambulatory leader may focus on inboxes, referrals, and preventive care gaps. An inpatient leader may work on orders, handoffs, and documentation. Population health work depends on registries, reliable measures, and coordinated outreach, and it may sit with a different team entirely.

At the CMIO or chief health information officer (CHIO) level, the remit extends across services and professions. You help set priorities, decide which requests get resources, and oversee clinical adoption and safety. These roles typically go to physicians with substantial informatics leadership experience, often as a prior CMIO. Board certification in clinical informatics is required by some employers and preferred by others, so read the actual requirements. Ask about decision rights, analyst support, and the clinical schedule. A bigger title without the resources to deliver is a hard job.

Industry

Industry lets you work on products used across many care settings, but the entry points differ. A short AI evaluation contract, an advisory relationship, and a full time product role demand different skills and build different experience.

Start with a bounded contribution

A growing number of platforms and recruiters connect physicians with paid expert work for AI companies: writing clinical scenarios and reference answers, evaluating model outputs, and grading against rubrics. It is a concrete way to see how clinical judgment shapes AI development. Describe it accurately as expert contribution or AI evaluation, not consulting or product leadership. Read the contract for hours, deliverables, payment terms, and ownership of the work, and expect that selection, continuity, and earnings are not guaranteed.

For advisory work, start with problems you know well. Innovation events, HIMSS communities, and physician entrepreneur networks can connect you with people building in that space. A useful first conversation might start with a workflow problem or an evaluation you could help design. Contributing to something specific lets both sides see how you work.

You may already know a digital health company from an implementation or pilot at your health system. Reconnecting about a non-confidential problem can be a reasonable step toward an advisory relationship. Check your institution’s outside work, conflict of interest, and intellectual property policies first, especially if you influence purchasing or evaluate that company’s product. Patient data and proprietary information do not become yours to share because a company values your expertise.

Move into a product or clinical leadership role

A clinical product role may involve translating a care problem into requirements, working with engineers and designers, and deciding whether the result is useful and safe. A clinical scientist role emphasizes evaluation and research, and some accept an MD as one of several qualifying degrees while expecting substantial research experience. Medical director and chief medical officer roles add responsibility for clinical services, protocols, physician teams, and quality. Titles overlap, so ask what decisions the role actually owns.

Joining a startup may mean trading some cash for an opportunity you value and an equity grant. It can also mean thinner benefits, shifting responsibilities, and funding uncertainty. Compare the actual offer rather than assuming every startup pays less. Ask about runway and what the team expects you to accomplish. For equity, understand the instrument, vesting, exercise costs, and what happens if you leave. Ownership can be diluted, and private company equity may never become spendable cash. Weigh it alongside salary, never as part of guaranteed pay.

Found your own company

Eventually you may want to build the company yourself. First test whether the clinical problem you see is one users and buyers will actually prioritize. Talk to them before committing to a solution, and find collaborators with the technical or commercial experience you lack. A pilot can establish feasibility, but you also need a plan for evidence, privacy, regulation, and who will pay. Resolve employer intellectual property and conflict questions early.

There is no dependable founder salary to promise. Early pay may be low or absent, and any eventual return depends on what the business can sustain and whether ownership ever turns into liquidity. Founding a company gives you direct responsibility for a solution you believe patients need. It also makes building the organization part of the job.

Keep patient care in the picture

Telehealth can be a full clinical career or a complement to other work. Video visits, messaging based care, and remote monitoring offer different ways to reach patients, and some digital health companies hire physicians for asynchronous care alongside opportunities to work with their engineers and designers. Before treating a remote role as an easy add to your week, ask about patient volume, response time expectations, follow up, and access to in person care. Check licensure requirements for the state where each patient is located.

What these roles pay

The ranges below are approximate U.S. figures drawn from published surveys and a sample of employer postings from 2025 and 2026. They are meant to orient you, not to benchmark an offer. Pay varies with region, organization size, your clinical specialty, and how much of the role is clinical. Always separate base salary, bonus, benefits, protected time, and equity when you compare offers, and note that an hourly contract rate is not equivalent to a salaried job with benefits.

Academic and health system roles

Informatics effort within a clinical job (physician informaticist, medical director, 0.1 to 0.5 FTE): Usually paid at or near your clinical rate, prorated to the informatics FTE; some organizations pay a fixed stipend instead. What drives it: Your specialty’s pay scale and whether the time is truly protected.

Academic faculty with an informatics focus: Generally follows the department’s specialty scale; roughly $150,000 to $300,000 for primary care based faculty. What drives it: Specialty, clinical share, and how much research or teaching is funded.

Fellowship program director: Typically 0.2 FTE or more of protected time rather than added pay. What drives it: Program size and institutional policy.

Research with a career development award: Grant covers part of your salary; the institution covers the rest. What drives it: Funding agency salary caps and departmental support.

Assistant or associate CMIO: Roughly $250,000 to $400,000 total, often with about half of the role clinical. What drives it: Organization size, scope, and specialty.

CMIO: Most commonly $350,000 to $500,000 base; about one in eight above $500,000; most are bonus eligible. What drives it: System size and scope of authority.

CHIO, chief health AI officer, CRIO, or similar enterprise role: Generally at or above the CMIO range; public data are limited. What drives it: Enterprise scope, reporting line, and whether the role is executive or academic.

Industry roles

AI evaluation and expert contract work: About $100 to $250 per hour; hours vary and benefits are not included. What drives it: Specialty, project demand, and contract terms.

Advisor to an early stage company: Often a small equity grant, sometimes paired with an hourly or retainer fee. What drives it: Stage of the company and the time commitment.

Clinical product manager, clinical staff, or clinical scientist: Roughly $170,000 to $300,000 base, usually plus equity. What drives it: Company stage, location, and research or product experience.

Medical director or VP of clinical at a digital health company: Roughly $250,000 to $375,000 base, plus equity and bonus. What drives it: Clinical operations scope and team size.

Chief medical officer at a venture backed company: Roughly $300,000 to $450,000 base plus equity; higher at larger or public companies. What drives it: Funding stage, company size, and regulatory exposure.

Asynchronous or virtual care contract physician: Roughly $100 to $200 per hour, or per encounter. What drives it: Specialty, volume, and licensure footprint.

Founder: No dependable range; early pay is often well below market or zero. What drives it: Funding, revenue, and eventual liquidity.

CMIO base salary distribution is from WittKieffer’s 2025 CMIO survey, as reported by Becker’s Hospital Review.

Where to begin

I came to informatics because I believed we could make care better. That can mean treating a patient directly, improving the system they rely on, teaching another physician, or building a tool worth using. Pick a problem you want to own, find people who can help you work on it, and ask how the work will be supported. That is a better beginning than waiting for the perfect title.

The views and opinions expressed in this post are my own and do not reflect the views of any organization, employer, or entity I work for or am affiliated with.

Next
Next

Stop Fast Forwarding: What Two Weeks in Europe Taught Me