Role Definition
| Field | Value |
|---|---|
| Job Title | Radiation Oncologist |
| Seniority Level | Mid-to-Senior |
| Primary Function | Evaluates cancer patients for radiation therapy candidacy, designs radiation treatment plans (dose, fractionation, technique selection), reviews AI-generated contours and treatment plans, manages on-treatment patients for acute toxicity and side effects, and follows patients post-treatment for recurrence and late effects. Participates in multidisciplinary tumor boards to determine optimal treatment sequencing alongside surgery and chemotherapy. |
| What This Role Is NOT | NOT a Radiation Therapist (technician who operates the linear accelerator). NOT a Medical Physicist (performs QA and dosimetry calculations). NOT a Medical Oncologist (prescribes chemotherapy). NOT a Diagnostic Radiologist (interprets imaging without treating). NOT a Medical Dosimetrist (creates dose distribution plans under physician direction). |
| Typical Experience | 12-20+ years post-graduation. MD/DO + 5-year radiation oncology residency + ABR board certification. Optional fellowship in brachytherapy, proton therapy, or pediatric radiation oncology. |
Seniority note: Junior residents in training would score similarly — the residency structure means independent practice begins at the mid-to-senior level. There is no meaningful "entry-level radiation oncologist" role.
Protective Principles + AI Growth Correlation
| Principle | Score (0-3) | Rationale |
|---|---|---|
| Embodied Physicality | 1 | Patient physical examination, brachytherapy implant procedures, and simulation positioning require hands-on contact. However, the majority of treatment planning work is digital. |
| Deep Interpersonal Connection | 2 | Cancer diagnosis discussions, treatment counseling, managing patient anxiety through multi-week treatment courses, palliative care conversations, and longitudinal survivor relationships. Trust is central to the therapeutic relationship. |
| Goal-Setting & Moral Judgment | 3 | Decides WHETHER to treat with radiation (vs surgery, chemo, or observation), WHAT dose and fractionation to prescribe, and balances curative intent against quality-of-life toxicity. Accountable for radiation-induced harm — second malignancies, organ damage, death. |
| Protective Total | 6/9 | |
| AI Growth Correlation | 0 | Cancer incidence — driven by aging populations — determines demand for radiation oncologists, not AI adoption. AI makes each physician more productive but does not create or eliminate the role. |
Quick screen result: Protective 6/9 — likely Green Zone (proceed to confirm).
Task Decomposition (Agentic AI Scoring)
| Task | Time % | Score (1-5) | Weighted | Aug/Disp | Rationale |
|---|---|---|---|---|---|
| Patient consultation & clinical assessment | 25% | 1 | 0.25 | NOT INVOLVED | Face-to-face cancer diagnosis discussions, physical examination, treatment intent decisions (curative vs palliative), informed consent. The human physician IS the value — no patient accepts an AI prescribing their radiation. |
| Treatment planning design & review | 25% | 3 | 0.75 | AUGMENTATION | AI auto-contouring (MVision, RayStation ML, Radformation) generates organ-at-risk and target volume contours; AI optimises dose distributions. GE iRT reduced sim-to-plan from 7 days to 7 minutes. Physician reviews, modifies, and approves — but AI handles significant sub-workflows. |
| On-treatment patient management | 20% | 2 | 0.40 | AUGMENTATION | Weekly on-treatment visits — assessing radiation dermatitis, mucositis, fatigue, pain. Adjusting supportive medications. AI can flag abnormal labs or predict toxicity risk, but the clinical assessment and medication decisions remain physician-led. |
| Multidisciplinary tumor board | 10% | 1 | 0.10 | NOT INVOLVED | Real-time collaborative decision-making with surgeons, medical oncologists, pathologists, and radiologists. Arguing treatment sequencing, weighing surgical margins against radiation fields, adapting plans to patient comorbidities. Irreducibly human deliberation. |
| Treatment plan approval & QA review | 10% | 2 | 0.20 | AUGMENTATION | Reviews dosimetric data, approves adaptive re-plans (Varian Ethos), ensures plan meets constraints. AI pre-checks flag deviations, but physician bears legal responsibility for every approved plan. |
| Documentation & administrative | 10% | 4 | 0.40 | DISPLACEMENT | Clinical notes, treatment summaries, insurance pre-authorisations, coding. DAX/Nuance ambient documentation and EHR AI modules handle the bulk. Physician reviews but does not generate from scratch. |
| Total | 100% | 2.10 |
Task Resistance Score: 6.00 - 2.10 = 3.90/5.0
Displacement/Augmentation split: 10% displacement, 55% augmentation, 35% not involved.
Reinstatement check (Acemoglu): Yes. AI creates new tasks — reviewing and validating AI-generated contours, interpreting AI-predicted toxicity models, overseeing adaptive radiotherapy workflows (Ethos re-plans during treatment), and integrating radiomics/genomics data into personalised dose prescriptions. The role is shifting from manual planning mechanics toward higher-order clinical decision-making.
Evidence Score
| Dimension | Score (-2 to 2) | Evidence |
|---|---|---|
| Job Posting Trends | 0 | ASTRO projects supply-demand balance through 2030. Declining residency applications since 2017 suggest stable rather than growing demand. Practice consolidation (large practices +51%, solo -27% between 2015-2023) reshapes the market but does not shrink it. |
| Company Actions | 0 | No AI-driven layoffs or headcount reductions reported. Practice consolidation is driven by economics and reimbursement, not AI displacement. AI tools are purchased as productivity enhancers, not physician replacements. |
| Wage Trends | 1 | Median compensation $400K-$472K+ with strong stability. Oncology salaries continue to grow, tracking or exceeding physician market averages. No wage compression signal from AI. |
| AI Tool Maturity | 0 | Production-deployed AI tools for auto-contouring (MVision, RayStation ML, Radformation, GE iRT) and adaptive planning (Varian Ethos) are widespread. However, all tools augment — physician reviews every contour and approves every plan. No tool prescribes dose or manages toxicity independently. |
| Expert Consensus | 1 | Broad agreement that AI transforms radiation oncology workflow (from manual contouring to clinical oversight) without displacing physicians. PMC consensus: AI shifts role from technical to clinical/holistic. No credible source predicts physician displacement. |
| Total | 2 |
Barrier Assessment
Reframed question: What prevents AI execution even when programmatically possible?
| Barrier | Score (0-2) | Rationale |
|---|---|---|
| Regulatory/Licensing | 2 | MD/DO degree + 5-year residency + ABR board certification + state medical license + DEA registration. No regulatory pathway exists for AI to independently prescribe radiation therapy. EU AI Act classifies radiotherapy AI as high-risk requiring human oversight. |
| Physical Presence | 1 | Patient examination, brachytherapy procedures, and simulation setup require physical presence. Follow-up visits increasingly via telemedicine, but initial consultations and procedures need hands-on contact. |
| Union/Collective Bargaining | 0 | Physicians generally not unionised in the US or UK. |
| Liability/Accountability | 2 | Radiation can cause severe injury or death — wrong dose, wrong site, wrong fractionation. Every treatment plan requires a physician signature. Personal malpractice liability cannot be transferred to AI. Someone must bear legal accountability for radiation-induced harm. |
| Cultural/Ethical | 2 | Cancer patients will not accept AI autonomously prescribing radiation to their bodies. The physician-patient relationship during cancer treatment is among the deepest trust relationships in medicine. Society demands a human physician making life-and-death radiation decisions. |
| Total | 7/10 |
AI Growth Correlation Check
Confirmed at 0 (Neutral). Cancer incidence is driven by demographics (aging populations), not AI adoption. AI makes each radiation oncologist more productive — auto-contouring compresses planning timelines dramatically — but the demand for the specialty is a function of how many people develop cancer, not how many AI systems are deployed. The ASTRO supply-demand balance projection through 2030 reflects this: stable demand, stable supply.
JobZone Composite Score (AIJRI)
| Input | Value |
|---|---|
| Task Resistance Score | 3.90/5.0 |
| Evidence Modifier | 1.0 + (2 × 0.04) = 1.08 |
| Barrier Modifier | 1.0 + (7 × 0.02) = 1.14 |
| Growth Modifier | 1.0 + (0 × 0.05) = 1.00 |
Raw: 3.90 × 1.08 × 1.14 × 1.00 = 4.8017
JobZone Score: (4.8017 - 0.54) / 7.93 × 100 = 53.7/100
Zone: GREEN (Green ≥48, Yellow 25-47, Red <25)
Sub-Label Determination
| Metric | Value |
|---|---|
| % of task time scoring 3+ | 35% (planning 25% + docs 10%) |
| AI Growth Correlation | 0 |
| Sub-label | Green (Transforming) — AIJRI ≥48 AND ≥20% of task time scores 3+ |
Assessor override: None — formula score accepted.
Assessor Commentary
Score vs Reality Check
The 53.7 score places this role comfortably in Green, and the label is honest. The score sits between Radiologist (52.7) and Oncologist (66.5), which is the correct calibration — radiation oncology combines the AI-exposed planning workflow of radiology with the treatment relationship of oncology. The 7/10 barriers are doing meaningful work here: strip them and the score drops to ~47 (Yellow boundary). This barrier dependency is justified — medical licensing, malpractice liability, and cultural trust in physician-directed cancer treatment are structural, not temporal. They are not eroding.
What the Numbers Don't Capture
- Productivity gains compressing headcount growth. GE iRT compresses sim-to-plan from 7 days to 7 minutes. If each radiation oncologist can plan and treat 30-50% more patients with AI assistance, the same cancer caseload requires fewer physicians. ASTRO projects balance through 2030, but this may mask a dynamic where productivity growth absorbs what would have been hiring growth.
- Declining residency interest. Five consecutive years of falling applications suggests the profession's perception among medical students is cooling. This could create a delayed shortage — fewer trainees entering now means fewer mid-career specialists in 10 years — which would paradoxically strengthen the position of those already in practice.
- Practice consolidation reshaping the market. Solo practices down 27%, large groups up 51%. The surviving radiation oncologist increasingly works in a large group or academic centre, not independent practice. This concentrates employment but does not eliminate positions.
Who Should Worry (and Who Shouldn't)
If you are a radiation oncologist who has embraced AI-assisted planning workflows — reviewing and modifying AI-generated contours rather than drawing them manually, using adaptive platforms like Ethos, and spending the time savings on patient care and tumor board participation — you are firmly Green. The AI makes you faster and more consistent, not redundant.
If you are a radiation oncologist whose primary identity is the technical craftsmanship of manual contouring and planning — the "artisan planner" who resists AI tools — you are not at displacement risk, but you are at competitive disadvantage. Practices will favour physicians who can see more patients with AI assistance over those who insist on manual workflows.
The single biggest separator is not AI adoption but patient-facing clinical skills. The radiation oncologist who spends liberated planning time on deeper patient counseling, tumor board leadership, and multidisciplinary coordination is the most protected. The one who spends it doing the same manual work AI could handle is the most vulnerable to productivity-based headcount compression.
What This Means
The role in 2028: The radiation oncologist of 2028 spends significantly less time on contouring and planning mechanics — AI handles first-pass contouring and plan optimisation in minutes. The physician's day shifts toward clinical assessment, toxicity management, adaptive treatment decisions, and patient counseling. The role becomes more clinical and less technical, which is a genuine improvement for patient care.
Survival strategy:
- Adopt AI planning tools fully. MVision, RayStation ML, Varian Ethos, GE iRT — master these workflows. The physician who validates AI contours in 10 minutes rather than drawing manually for 2 hours treats more patients and is more valuable to any practice.
- Invest in clinical and interpersonal skills. As AI absorbs planning mechanics, the differentiator becomes patient communication, palliative care expertise, and multidisciplinary leadership. These are the irreducible human tasks.
- Develop subspecialty expertise. Brachytherapy (hands-on procedures AI cannot perform), proton therapy, pediatric radiation oncology, and radiomics/precision medicine are areas where deep specialisation creates additional protection.
Timeline: 5-10+ years of Green stability. AI is transforming daily workflow substantially but is not approaching physician displacement. Barriers are structural and durable.
