Medical Workforce Capacity Model to support top management decisions in healthcare

Summary

The Russian healthcare system faces not merely a question of the current number of physicians.

A more fundamental question is this:

Is the healthcare system capable of independently reproducing sufficient medical capacity amidst demographic shifts in the population, the aging of the physician corps, the retirement of senior generations, changes in the attractiveness of medical education, and the rise of alternatives for young specialists?

This question is fundamentally different from "How many doctors does Russia need?"

The number of physicians is only one variable. For system sustainability, the following factors are simultaneously important:

• physician numbers;
• age structure;
• specialty;
• region;
• actual employment;
• professional longevity;
• influx of new specialists;
• residency completion;
• retention of physicians in the profession;
• working conditions;
• income relative to alternative professions;
• productivity;
• digitalization;
• preservation of institutional medical experience.

In 2026, signals emerged that warrant separate analysis.

On one hand, the Ministry of Health reports that for 20,000 budget‑funded residency positions, 39,000 applications were submitted, and therefore considers it premature to speak of a systemic shortfall.

On the other hand, major medical universities reported hundreds of unfilled budget‑funded residency positions and announced additional enrollment. As of mid‑August, vacancies were reported at Pirogov Russian National Research Medical University, Sechenov First Moscow State Medical University, the Russian Medical Academy of Continuing Professional Education, and other institutions.

These observations do not contradict each other. Overall competition may remain high, while simultaneously individual regions, specialties, or targeted positions may experience candidate shortages.

This is precisely why what is needed is not a single competition statistic, but a dynamic model of medical workforce reproduction.

It is proposed to create a Medical Workforce Capacity Model (MWCM).

Its task is not to provide a "precise forecast through 2045," but to determine:

• which parameters determine the sustainability of the medical system;
• which of them are most sensitive;
• where shortages emerge first;
• what effect each individual parameter change produces;
• what the delay is between a policy decision and its result;
• which measures compensate for demographic decline;
• where the boundary lies between temporary shortage and self‑sustaining workforce crisis.

What the Model Does NOT Do

To avoid misinterpretation, it is essential to state clearly what MWCM is not:

• It is not a demographic forecast of the total number of physicians in 2045.
• It is not a detailed workforce plan for the entire country.
• It is not a justification for indiscriminately increasing or decreasing budget‑funded places without considering specialty and region.
• It is not a substitute for clinical expertise or evidence‑based medicine.
• It is not an excuse to lower training standards or quality of care.
• It is not a crystal ball — it is a tool for exploring dependencies, comparing scenarios, and identifying leverage points.

The model is useful precisely because it makes visible the trade‑offs and time lags that are otherwise hidden in aggregated statistics.

1. The Problem

Traditional workforce planning often relies on metrics such as:

• physicians per 10,000 population;
• number of graduates;
• number of budget‑funded places;
• number of vacancies;
• number of applications;
• average physician age.

These indicators are necessary but insufficient.

For example, a system can simultaneously have a large number of students, high competition for medical schools, physician shortages in regions, an excess of candidates for some specialties, unfilled positions in others, a large number of older physicians, growing workload, and increasing healthcare expenditures.

This is not a contradiction. It is the result of different parts of the system moving at different speeds.

2. Core Idea of the Model

It is proposed to distinguish between the number of physicians \(D_t\) and effective medical capacity \(EC_t\).

Effective capacity is determined not only by the number of physicians:

\[ EC_t = D_t \times A_t \times P_t \]

where:

• \(D_t\) — number of physicians;
• \(A_t\) — actual availability of physicians for clinical work (share of working time spent on direct patient care);
• \(P_t\) — productivity per physician (e.g., number of consultations or procedures per unit of clinical time).

A 10% increase in the number of physicians does not necessarily mean a 10% increase in medical capacity. If at the same time bureaucratic burden has grown, actual employment has decreased, or patients have become more complex, the effect may be substantially smaller. Conversely, automation of part of the work may increase effective capacity with an almost unchanged number of physicians.

3. Reproduction of the Physician Corps

The number of physicians changes as a stock affected by flows:

\[ D_{t+1} = D_t + N_t + I_t - R_t - L_t \]

where:

• \(N_t\) — new physicians entering practice;
• \(I_t\) — immigration inflow;
• \(R_t\) — exit from the profession due to age;
• \(L_t\) — other losses: leaving clinical medicine, career change, emigration, transition to administrative roles, etc.

This yields the first key indicator — the physician reproduction rate:

\[ RR_t = \frac{N_t + I_t}{R_t + L_t} \]

Interpretation:

• \(RR > 1\) — generation is reproducing;
• \(RR = 1\) — approximate equilibrium;
• \(RR < 1\) — the physician stock begins to structurally contract.

What matters is not only the national \(RR\), but also \(RR\) by region and specialty.

4. Why Generational Change Is a Separate Problem

The physician corps has an age structure. Therefore, it is necessary to model age groups: 25–29, 30–34, 35–39, 40–44, 45–49, 50–54, 55–59, 60–64, 65–69, 70+.

This allows us to see not only the current shortage, but also the future wave of attrition. Particularly important is the situation where the younger generation is smaller than the older generation approaching retirement. In this case, the problem becomes partially predetermined.

4.5. Regional Heterogeneity as a Separate Risk

National averages mask deep regional disparities. Even if the national \(RR = 1\), individual regions may have \(RR = 0.6\) while others have \(RR = 1.4\).

• National aggregates hide crisis points;
• Physician mobility between regions is limited;
• Solutions must be tailored to specific regional and specialty contexts.

Therefore, the model should output a map of \(RR\) by region and specialty, not a single national number.

5. Loss of Experience — A Separate Component of Risk

The departure of a physician means the loss of not just one working unit. In some cases, along with the physician, the system loses accumulated clinical experience, the ability to recognize rare cases, knowledge of local processes, mentoring skills, informal problem‑solving approaches, and institutional memory.

Therefore, capacity loss ≠ numerical loss. In some specialties, capacity loss significantly exceeds numerical loss.

This is precisely where this model connects with the previously proposed concept of Digital Memory of Medical Institutions — the preservation and reuse of accumulated clinical experience. The concept envisions creating an infrastructure layer atop existing medical information systems that transforms documented practice into structured, traceable, and reusable institutional memory.

6. The Educational Pipeline

A physician does not appear in the system immediately after the state decides to increase the number of budget‑funded places. Simplistically:

\[ S_t \rightarrow M_t \rightarrow G_t \rightarrow O_t \rightarrow N_t \]

where \(S_t\) — potential applicant pool; \(M_t\) — admitted; \(G_t\) — graduates; \(O_t\) — those completing required specialization; \(N_t\) — entering practice.

Thus, a budget‑funded place ≠ a future physician. There are several filters between them.

7. A Young Person's Choice of Medicine

The number of potential physicians depends not only on demography. Simplistically:

\[ N_t = S_t \times C_t \times G_t \times O_t \times R_t \]

where \(C_t\) — probability of choosing a medical career. It depends on expected income, duration of training, working conditions, professional autonomy, prestige, legal liability, emotional burden, career advancement opportunities, geographic mobility, residency conditions, and alternative careers.

8. The Price of Alternatives

For a young person, medicine does not compete with "nothing." It competes with IT, engineering, finance, entrepreneurship, management, applied science, and other high‑skilled professions.

Therefore, it is necessary to account for the opportunity cost of choosing medicine (\(OC\)): alternative income + flexibility + mobility + speed to income + career opportunities.

Even a high absolute physician salary may be insufficient if the relative attractiveness of alternatives is growing faster.

9. Working Conditions

The attractiveness of the profession is determined not only by income. Simplistically:

\[ MA = f(\text{Income}, \text{Autonomy}, \text{Conditions}, \text{Prestige}, \text{Mobility}, \text{Technology}, \text{Training Duration}, \text{Legal Risk}, \text{Alternatives}) \]

What is particularly important is that some parameters affect multiple stages simultaneously: working conditions → admission → residency → entry into the profession → retention.

Therefore, improving working conditions may have a greater systemic effect than simply increasing the number of budget‑funded places.

10. Medical Demand

It is also necessary to model the other side of the system:

\[ H_t = \text{Population}_t \times \text{AgeIntensity}_t \times \text{Morbidity}_t \times \text{MedicalIntensity}_t \]

where Population — population size; AgeIntensity — intensity of medical care depending on age; Morbidity — disease structure; MedicalIntensity — intensity of medical technology use.

Therefore, a decline in population does not necessarily mean a decline in demand for medical care. Population aging may compensate for or exceed the effect of population decline.

11. Technology as a Multiplier of Workforce Capacity

In the model, technology should not be viewed solely as a way to "replace a physician." It is more useful to view it as a multiplier \(EC_t = D_t \times A_t \times P_t\), where \(P_t\) can increase through documentation automation, AI, clinical decision support systems, telemedicine, automation of routine operations, improved information exchange, task redistribution, and digital memory of the medical organization.

11.5. Quality as a Multiplier

Increasing the number of physicians does not automatically preserve or improve quality. If numerical growth is achieved by lowering admission standards, reducing clinical training, or overburdening mentors, the effective capacity may actually decline.

Therefore, we introduce a quality coefficient \(Q_t\) — the fraction of physicians who meet clinical competency standards.

\[ EC_t = D_t \times A_t \times P_t \times Q_t \]

\(Q_t\) can be estimated via certification results, peer reviews, or patient outcomes. It ensures that policy makers do not equate "more doctors" with "better care."

12. Digital Memory as an Element of Workforce Sustainability

If a specialist's practical experience is preserved only in their memory, then the specialist's departure leads to a loss of experience. If the experience is structured and accessible to the next generation, the specialist's departure leads to partial transfer of experience.

Therefore: \(\text{KnowledgeRetention} \rightarrow \text{Productivity} \rightarrow \text{EffectiveCapacity}\).

At the same time, digital memory should not replace the physician, clinical guidelines, evidence‑based medicine, or medical responsibility. Documented practice must be distinguished from proven causal effect.

13. Controllable Parameters of the Model

The main practical purpose of the model is not forecasting for its own sake. It should answer the question: which parameter must be changed to obtain a specific result?

Key levers:

Parameter	What We Change	Direct Effect	Delay	Type
N	number of new physicians	increases workforce stock	high	slow
R	retirement age / conditions	retains physicians	low–medium	fast/medium
L	departure from profession	reduces losses	medium	medium
A	clinical work time share	increases available capacity	low	fast
P	productivity	increases effective capacity	low–medium	fast/medium
C	medical attractiveness	increases future inflow	high	slow
OC	alternative attractiveness	reduces future inflow	high	slow
K	knowledge retention	reduces knowledge loss	medium	medium
I	immigration	rapidly increases stock	low	fast
H	healthcare organization	changes required capacity	medium	medium
Q	quality standards	increases effective capacity	medium	medium

Note that \(Q\) is now an explicit lever, reinforcing that quality is not automatic.

14. How the Model Can Be Used for Management

This is a fundamental part of the model. The goal is to turn analysis into a decision‑making tool of the form: goal → parameter → magnitude of change → delay → expected effect → side effects.

14.1. Goal 1. Rapidly Increase Available Medical Capacity

The fastest levers: \(A\uparrow\) and \(P\uparrow\). That is, reducing bureaucracy, automating documentation, redistributing tasks, improving scheduling, eliminating downtime, digital tools.

Advantage: results can appear significantly sooner than a new physician completes training.

14.2. Goal 2. Reduce Future Workforce Shortages

The main lever is \(N\uparrow\). But \(N\) can be increased in several ways:

• Option A: increase the number of students → effect through the duration of the educational cycle;
• Option B: increase the proportion of graduates who continue training (\(O\uparrow\));
• Option C: increase the proportion of graduates who remain in the profession (\(Retention\uparrow\)).

In most cases, B and C may be cheaper than endlessly increasing the input flow.

14.3. Goal 3. Stop Generational Decline

The main indicator: \(RR = \text{New Physicians} / \text{Departing Physicians}\). To achieve \(RR \ge 1\), both sides can be influenced.

Increase the numerator:

• increase input;
• improve residency;
• improve working conditions;
• increase retention of young specialists;
• attract immigration.

Decrease the denominator:

• make continued work by senior physicians more attractive;
• reduce administrative burden;
• create part‑time employment;
• use the experience of senior physicians as a mentoring resource.

This is far more rational than trying to solve everything solely by increasing enrollment.

14.4. Goal 4. Reduce Regional Shortages

Here, changing the national \(D\) may achieve almost nothing. It is necessary to influence \(RR\) by region and specialty.

Main levers: regional salary, housing, career trajectory, educational infrastructure, possibility of returning to a major center after mandatory service, remote support, access to expert experience from leading centers.

If a physician is in a region but gains access to institutional experience from a leading center, the effective capacity of the regional system can grow without the physical relocation of the physician.

14.5. Goal 5. Maintain Quality as the Physician Corps Ages

Main parameter: \(K = \text{KnowledgeRetention}\). It can be enhanced through mentoring, structured experience transfer, digital memory, case‑based learning, documentation of complex cases, connecting leading centers with peripheral institutions.

Additionally, \(Q\) must be monitored — older physicians may have deep experience but also may face cognitive decline or outdated practices. The model should allow for differential \(Q\) by age group.

14.6. Goal 6. Increase Effective Capacity Without Proportional Growth in Numbers

This is the most technological path: \(P\uparrow\). If the number of physicians is unchanged (\(D = const\)), but \(A\uparrow\) or \(P\uparrow\), then \(EC\uparrow\).

This is precisely why automation, AI, and work organization improvements should be viewed not as a separate digital program, but as part of workforce policy.

15. Model Sensitivity and a Realistic Example

For a first approximation, it is convenient to use relative changes. If \(EC = D \times A \times P \times Q\), then for small changes:

\[ \frac{\Delta EC}{EC} \approx \frac{\Delta D}{D} + \frac{\Delta A}{A} + \frac{\Delta P}{P} + \frac{\Delta Q}{Q} \]

This means that a 10% increase in effective capacity can be achieved through various combinations of smaller improvements.

Illustrative Example Based on Real Data

We use data from a photo‑timing study of abdominal surgeons at City Clinical Hospital No. 6 in Penza (RMANPO, 2024–2026). The study measured actual working time allocation.

• Direct patient care (clinical work): 62.3% for daytime surgeons, 67.8% for on‑call surgeons.
• Documentation and administrative tasks: ~9–12%.
• Auxiliary processes (transfers, waiting): ~18–20%.

For our example, we take a conservative \(A = 0.62\) (62% of working time spent on direct clinical activity). This is a midpoint between daytime and on‑call values. Note that this value applies to this specific group; for other specialties or regions, \(A\) may differ significantly (e.g., primary care physicians may have lower \(A\) due to higher documentation burden).

Assume a department with \(D = 100\) surgeons, \(P = 1.0\) (baseline productivity), and \(Q = 1.0\) (all meet quality standards).

Current effective capacity: \[ EC = 100 \times 0.62 \times 1.0 \times 1.0 = 62 \]

To increase \(EC\) by 10% (to 68.2), we compare three strategies:

Strategy	Change	New EC	Time to effect
A (reduce bureaucracy)	\(A: 0.62 \to 0.682\)	68.2	months
P (boost productivity)	\(P: 1.0 \to 1.1\)	68.2	1–3 years
D (hire more surgeons)	\(D: 100 \to 110\)	68.2	7+ years

All three yield the same gain, but the time horizons are drastically different. A combined approach (\(D+3.3\%,\ A+3.3\%,\ P+3.3\%\)) also gives ~10.2% gain (\(1.033^3 \approx 1.102\)) with moderate efforts on each front.

Important caveat: The \(A=0.62\) value is specific to the studied surgeons. In regions with higher administrative burden (as reported by the Audit Chamber, up to 50% of working time lost to non‑clinical tasks), \(A\) could be as low as 0.50. Therefore, the model must be calibrated locally.

Also, the study did not capture 36‑hour on‑call shifts (which exist in practice but were not measured). Such extended shifts may reduce effective \(A\) due to fatigue, but they are not reflected in the data. This underlines the need for further empirical work.

16. Why This Changes the Policy Approach

Suppose it is necessary to increase effective capacity by 15%. There are four fundamentally different strategies.

Strategy 1 — education only: \(D +15\%\). Results appear in years.

Strategy 2 — productivity only: \(P +15\%\). Results may appear significantly faster.

Strategy 3 — work organization: \(A +15\%\). Part of the effect may be achieved even faster.

Strategy 4 — combined: \(D +5\%,\ A +5\%,\ P +5\%\).

In first approximation: \(1.05 \times 1.05 \times 1.05 \approx 1.158\) — that is, about 15.8% additional effective capacity.

This is not a forecast of a specific medical effect. It is a demonstration of the model's principle: several moderate changes can yield greater results than attempting to radically change a single parameter.

17. The Most Important Constraint: Delay

Parameters have different speeds of impact.

Fast:

• work organization;
• reduction of administrative burden;
• automation;
• FTE changes;
• part of immigration decisions.

Medium:

• physician retention;
• workload redistribution;
• mentoring;
• digital memory.

Slow:

• change in professional attractiveness;
• change in educational flow;
• training of new specialists;
• demographic structure.

Therefore, a rational strategy must use a portfolio of levers with different time horizons.

17.5. Sustainability Buffer

A simple but powerful derived indicator is the buffer \(B_t\):

\[ B_t = \frac{D_t - D_{min}}{D_{min}} \]

where \(D_{min}\) is the minimum number of physicians required to provide essential care (estimated from demand models).

• \(B < 0.1\) — the system is in the danger zone;
• \(B < 0\) — critical deficit.

This helps distinguish between "shortage" and "crisis" and provides an early warning signal that is easy to communicate.

18. Scenarios 2026–2045

The scenarios presented are purely illustrative and serve to demonstrate the system's sensitivity to parameter changes. They are not forecasts and do not claim quantitative accuracy.

Scenario A — Demographic Stress. Assumes a decline in the number of young cohorts, high physician workload, continued retirement of older generations, weak improvement in conditions, low productivity growth. Result: \(RR < 1\) over a prolonged interval. The system enters self‑sustaining deficit.

Scenario B — Inertial. Gradual increase in training, moderate salary increases, moderate digitalization, maintaining the current system structure. Result: shortage does not necessarily become catastrophic, but remains chronic. This is the most dangerous scenario due to its external appearance of stability.

Scenario C — Improved Attractiveness. Changes: income↑, autonomy↑, conditions↑, administrative burden↓, career flexibility↑. Result appears with delay, since first the young person's choice changes, then education, then residency, then entry into the profession.

Scenario D — Productivity First. Physician numbers grow moderately. But \(P\uparrow\) through automation, AI, clinical information systems, task redistribution, digital memory, process improvement. This scenario allows partial compensation of demographic pressure without proportional growth in physician numbers.

Scenario E — Systemic Transformation. Simultaneously \(N\uparrow,\ Retention\uparrow,\ A\uparrow,\ P\uparrow,\ KnowledgeRetention\uparrow\), and targeted management of Region × Specialty × Age. This is the most sustainable scenario.

19. What the Model Should Actually Show Decision‑Makers

The output should answer not "In 2045 there will be 720 thousand physicians," but:

• What medical capacity is needed?
• What capacity will we obtain under current policy?
• Where does the gap arise?
• Which parameter is the main constraint?
• What happens if we change only this parameter?
• In how many years will the effect appear?
• What is the cost of changing the parameter?
• What risks arise when changing it?

This transforms the model from an analytical report into a decision‑making tool.

20. Recommended Early Warning Indicators

It is proposed to regularly track:

• Reproduction rate \(RR = \text{New Physicians} / \text{Departing Physicians}\);
• Transition rate from graduates to practice \(TCR\);
• Retention rate \(R_5\);
• Age risk — proportion of physicians above a given threshold;
• Clinical employment share \(A = \text{Clinical time} / \text{Total working time}\);
• Productivity \(P\);
• Quality coefficient \(Q\);
• Buffer \(B\);
• Knowledge loss index — proportion of critical competencies associated with physicians approaching retirement;
• Regional \(RR\) for each region‑specialty combination.

21. What Decisions the Model Allows Comparison Of

The model should allow questions such as:

• What gives more effect: +10% budget‑funded places or reducing administrative burden by 10%?
• What is more effective: increasing salary or increasing professional longevity?
• What effect will automating 20% of administrative work have?
• How many additional physicians can we avoid training if productivity grows by 5%?
• Which specialties require enrollment increases, and which primarily require condition improvements?
• What happens if the retirement of the older generation accelerates by five years?
• How will the system change if retention of young physicians increases by 10 percentage points?

These are precisely the questions that have direct value for state policy.

22. Model Limitations

22.1. "All Models Are Wrong, but Some Are Useful"

Any model is a simplification of reality. Therefore, "All models are wrong, but some are useful" is not a decorative caveat, but a principle for interpreting this document.

MWCM should not be used as a machine that produces the "true number of physicians for 2045." Its purpose is different:

• identify dependencies;
• compare scenarios;
• detect sensitive parameters;
• identify early signals;
• assess delays;
• test the robustness of decisions.

If data or assumptions change, the model results should change with them.

22.2. Data Limitations

A full‑fledged model requires data not presented in this conceptual document in a unified form: physician age, specialty, region, actual FTE, departure from the profession, transition between public and private sectors, graduate and resident trajectories, actual workload, productivity, medical demand by age groups.

Therefore, numerical scenarios should be viewed as sensitivity scenarios, not as an official demographic forecast.

22.3. Causality Limitation

Correlation does not imply causation. For example, salary growth does not guarantee increased enrollment because working conditions, prestige, alternative professions, educational requirements, and regional mobility may change simultaneously. Therefore, model parameters must be empirically validated.

22.4. Generality of the Example

The illustrative example in Section 15 is based on a specific study of abdominal surgeons in Penza. It should not be extrapolated to other specialties or regions without local calibration. The model is designed to be adaptable, but the numbers are not universal.

23. Author's Limitations

The author is not a physician, health economist, or specialist in Russian healthcare organization. Therefore, this document is not an expert medical opinion.

Nevertheless, the author has experience directly useful for framing the proposed task.

Author's Strengths

The author has worked for over 15 years as an independent software engineer, involved in software systems development, system architecture, R&D, complex system modeling, information security, analytical systems construction, information systems integration, knowledge management, AI application and automation, and research on the relationship between technology and productivity.

The author's previous work "How many software engineers the economy could support?" addressed a similar framing: not simply estimating the number of specialists, but modeling the economic capacity of a system to support a certain number of high‑skilled workers, accounting for productivity, feedback, implementation time, and other parameters.

That work also uses the principle that changing one system parameter can substantially change its steady state. This approach is transferred to the present model.

Missing Experience

The author lacks practical experience as a physician, clinical practice, healthcare facility management experience, access to restricted healthcare workforce statistics, or full‑scale expertise in health economics.

Therefore, medical specialists, demographers, health economists, and representatives of regional healthcare systems should be considered mandatory participants in the next stage of model development.

The strength of the present work is the systemic problem framing and model architecture, not a claim that the author possesses medical expertise.

24. Related Research by the Author

Below are publications directly related to the methodology and problem framing of the present model.

1. How many software engineers the economy could support?
October 2025. This work is the direct methodological predecessor of the present model. It examines base economic capacity, supporting population, productivity, feedback, implementation time, success of product creation, exports, automation, and the possibility of expanding the natural limit of the system through productivity growth.

Link: https://gelassen.github.io/blog/2025/10/09/how-many-sofware-engineers-the-economy-could-support.html

2. Bridging gaps in economy, politics, diplomacy and warfare: review and next steps, Part II
February 2026. This work describes the development of the author's systems approach to complex system analysis: network dependencies, multi‑level causal relationships, scenario thinking, and working with uncertainty. These methods are used in the present document when constructing the medical workforce reproduction model.

Link: https://gelassen.github.io/blog/2026/02/03/bridging-gaps-in-economy-politics-diplomacy-warfare-review-part-II.html

3. Digital Memory of Medical Institutions: Turning Clinical Experience into a System Asset
August 2026. This work examines the creation of digital memory of medical institutions as an infrastructure layer that transforms accumulated clinical practice into structured and traceable institutional knowledge. In the context of the workforce model, this can be viewed as one mechanism: KnowledgeRetention → Productivity → EffectiveCapacity.

Link: https://gelassen.github.io/blog/2026/08/18/digital-memory-of-medical-institutions.html

This set of publications demonstrates a consistent methodological line: from modeling the capacity of a system for high‑skilled human capital — through systems and scenario analysis — to applying these methods directly to healthcare and preservation of medical experience.

It is important to emphasize: medical expertise should come from the Ministry of Health, physicians, demographers, and health economists. The author's competence is system architecture, modeling, technology, and complex system analysis.

25. Proposed Pilot

To transform the concept into a working tool, it is proposed to begin not by building a massive state information system, but with an analytical pilot.

The pilot should include:

• 3–5 regions with different demographic situations;
• 5–10 specialties;
• several large medical organizations;
• medical universities;
• residency programs;
• data on physician age structure.

It is desirable to reconstruct dynamics for at least 2015–2026 and build scenarios for 2026–2045.

26. What the First Prototype Should Deliver

The first prototype should answer at least five questions:

• Where does \(RR < 1\) already exist?
• Which specialties are most sensitive to the departure of the older generation?
• Which regions will be unable to reproduce their own workforce stock?
• Which parameter has the greatest impact on effective capacity?
• Which measures yield results most quickly?

Concrete pilot question

"In Sverdlovsk Oblast, for the surgery specialty, 40% of surgeons are older than 55, and residency enrollment is only 60% of the target. What is the effect of increasing retention of young surgeons by 10 percentage points compared to increasing enrollment by 20%?"

The model should be able to simulate both scenarios and compare them in terms of time, cost, and impact on \(RR\) and \(EC\).

The pilot is considered successful if it allows ranking 3–5 specific measures by the ratio of "effect / lag / cost" for selected specialties and regions.

27. Desired Outcome

The final result should not be a single number. What is needed is a map of workforce sustainability: Region × Specialty × Generation × Scenario.

Such a format is far more useful than a single indicator like "Russia is short by N physicians."

28. Main Conclusion

The Russian healthcare workforce problem should be viewed not as a task to "train more physicians."

A more precise formulation is: "ensure the reproduction of sufficient effective medical capacity amidst demographic changes in the population and physician generations."

To achieve this, it is necessary to simultaneously manage: quantity + age + specialty + geography + retention + conditions + productivity + technology + quality + knowledge.

29. Strategic Principle

If a shortage arises today, a solution through increasing student numbers may come too late.

If the older generation is retiring today, training a new specialist will not compensate for it immediately.

If a physician spends a significant portion of working time not on clinical activity, increasing headcount may be less effective than eliminating organizational losses.

If unique experience disappears with the physician, a new diploma alone is insufficient to fully replace the lost capability of the system.

Therefore, workforce policy must simultaneously operate on three horizons:

• Today: \(A\uparrow,\ P\uparrow\) — increase the effective capacity of the existing corps.
• 3–7 years: \(Retention\uparrow,\ KnowledgeRetention\uparrow,\ Q\uparrow\) — preserve people, quality, and experience.
• 7–15+ years: \(N\uparrow,\ RR \ge 1\) — ensure generational reproduction.

30. Conclusion

The most dangerous mistake would be to wait until statistics show an obvious nationwide physician shortage. By that time, a significant part of the process will already be irreversible: the younger generation will have chosen other professions, the age structure of the physician corps will have shifted, specialists will have retired, and the educational cycle will no longer allow rapid replenishment of the deficit.

Therefore, a transition from reactive workforce policy to an early warning system is necessary.

The proposed model does not claim to be able to precisely predict the state of Russian healthcare in 2045. It proposes something else: to make visible the parameters on which that future depends.

If the model shows that to achieve a specific goal it is sufficient to change a certain parameter by a certain magnitude, this creates a foundation for rational choice between alternative state measures.

This is precisely why the ultimate goal should not be maximizing the number of physicians.

Goal: maximize sustainable effective medical capacity

while maintaining quality of care, patient safety, professional responsibility of physicians, accessibility of care, institutional memory, and the system's ability to reproduce itself across generations.

It is proposed to consider the Medical Workforce Capacity Model as a conceptual foundation for further joint research with medical specialists, demographers, health economists, educational organizations, and regional healthcare authorities.