Skip to sources
Time Machine

Medical Scientists, Except Epidemiologists

Scrub through 138years of this role's history, from when it first emerged, through every wave of technology that reshaped it, to the cited projections for where it's heading next.

2026drag to travel through time
19001925195019752000now
2026
Known today as Medical Scientist, Except Epidemiologist (BLS SOC 19-1042)
US Employment
172K
OEWS is a point-in-time survey snapshot, not a continuous time series; BLS advises against using it for year-over-year trend comparison.
Median Annual Wage
$103,410
≈ $100,759 in 2024 dollars
Each dot is a cited figure over time; the dotted line only links them (values between aren't measured). Hollow dots are estimates.
Tools of the era

The tools that defined the work

Select an era to see how it reshaped the work.

  • Compound microscope + bacterial culture (Koch's postulates era)

    Robert Koch's 1876 demonstration that Bacillus anthracis caused anthrax — proved by isolating, culturing, and reinjecting the organism — established the fundamental method of laboratory medicine: grow the agent, study it in isolation, prove causation by fulfilling postulates. The tools were the compound microscope (Zeiss objective lenses, by the 1880s resolving to ~0.2 microns), glass culture plates (Koch's gelatin medium, later agar), and Bunsen burners for sterile technique. American medical researchers trained in German laboratories brought these methods home. Koch-trained William Welch built the first US bacteriology laboratory at Johns Hopkins in 1884; his student Simon Flexner later directed the Rockefeller Institute (1903). The microscope and the petri dish were not merely tools — they defined what a medical scientist did. Every experiment for the next 60 years ultimately came back to something seen under a lens or grown on a plate.

    Effect on the work

    The Koch-Pasteur bacteriological revolution created the occupation. Before 1876, "medical research" meant anatomy, physiology, and clinical observation. After 1876, it meant microbiology — a laboratory discipline requiring trained personnel. The US population of medical laboratory scientists doubled each decade from 1890 to 1940.

    Work toolChanging equipment
  • Ultracentrifuge + electron microscope + radioisotope tracers

    Three instruments transformed medical research in the 1940s-1950s, each enabling a different scale of investigation. The ultracentrifuge (Svedberg, 1920s; analytical ultracentrifuge by 1940) allowed researchers to separate and characterize macromolecules by molecular weight — the first tool that could distinguish protein complexes and viruses by their physical properties. The electron microscope (commercial Siemens and RCA models by 1940) could resolve structures 1,000× smaller than the light microscope: virus particles, cell organelles, and eventually DNA. Radioisotope tracers (carbon-14 became widely available after WWII from Oak Ridge National Laboratory) allowed researchers to label metabolic compounds and follow them through biological processes — the method used by Calvin to map photosynthesis (1954) and by Kornberg to trace DNA replication (1956). These tools did not merely improve existing experiments; they opened entirely new questions that defined the biological revolution of the 1950s-1970s.

    Work toolChanging equipment
  • Molecular biology — DNA double helix + Watson-Crick base pairing

    Watson and Crick's April 25, 1953 Nature paper "A Structure for Deoxyribose Nucleic Acid" — 900 words and one diagram — reoriented all of medical research. Once the information-storage mechanism of life was understood to be a sequence of base pairs in a double helix, the program for the next 50 years of research became obvious: determine the sequence, understand the code, and intervene in disease at the level of DNA. The molecular biology revolution that followed created new research techniques at a remarkable rate: Meselson-Stahl DNA replication (1958); mRNA discovery (Jacob and Monod, 1961); the genetic code cracked (Nirenberg and Khorana, 1966 Nobel); restriction enzymes (Arber, Smith, Nathans — 1970 Nobel in 1978). Each advance created a new class of experiments that medical scientists could run. The paradigm shift was as profound as Koch's: before Watson-Crick, medical research studied organisms; after, it increasingly studied molecules.

    Work toolChanging equipment
  • Recombinant DNA + monoclonal antibodies + PCR

    Three techniques in quick succession gave medical scientists the ability to engineer biology rather than merely observe it. Herbert Boyer and Stanley Cohen's 1973 experiment cut and spliced DNA from two different organisms using restriction enzymes — the first recombinant DNA experiment. Three years later, Boyer co-founded Genentech, the first company built entirely on recombinant DNA, which produced recombinant human insulin in 1982 — the first recombinant pharmaceutical approved for human use. Köhler and Milstein's 1975 hybridoma technique (1984 Nobel) produced monoclonal antibodies — proteins precisely targeted to a single molecular epitope, which became the foundation of modern immunotherapy. Kary Mullis's polymerase chain reaction (PCR, 1983, 1993 Nobel) amplified trace quantities of DNA a billion-fold, making DNA analysis routine and launching diagnostics, forensics, and evolutionary biology as population-scale sciences. Together these three tools turned the hypothetical promises of molecular biology into commercial biotechnology, and created the pharmaceutical and biotech research employment that now accounts for over a third of the 19-1042 workforce.

    Effect on the work

    Genentech's 1980 IPO at $35/share (double the offering price within minutes) launched the US biotech industry. By 1990 there were over 1,300 US biotech companies; by 2000 over 1,500. Each required medical scientists. The biotech sector transformed the employment geography of the occupation: from almost entirely academic + government pre-1973, to roughly one-third industry by 2000.

    Work toolChanging equipment
  • Human Genome Project → next-generation sequencing

    The Human Genome Project, launched October 1990 as an international collaboration co-led by the NIH and the DOE, set out to sequence all 3.2 billion base pairs of the human genome. It completed its first working draft in June 2000, declared completion in April 2003, and at an eventual cost of approximately $3 billion established the reference sequence against which all subsequent genomics research is calibrated. The HGP's real legacy was not the sequence itself — it was the bioinformatics and sequencing-technology infrastructure built to produce it. Next-generation sequencing (Solexa / Illumina platform, commercially available from 2007) reduced the cost of a whole human genome from the HGP's $300M draft to under $1,500 by 2015 and under $200 by 2023. That 6-orders-of-magnitude cost collapse is the fastest sustained decline in the cost of any scientific instrument in history, and it created an entirely new sub-discipline of medical scientists — computational biologists, bioinformaticians, and genomic epidemiologists — working at the intersection of biology and data science.

    Effect on the work

    The NIH budget grew from $13.6B (2000) to $27.1B (2003) during the HGP completion and post-genomics expansion — a doubling in three years. This drove a commensurate expansion in funded research positions. The sequencing-cost collapse created a second wave of hiring as genomic analysis became accessible to every medical school, not just the five major sequencing centers.

    Work toolChanging equipment
  • AlphaFold v1→v2→v3 — protein-structure prediction at scale

    Protein folding had been the central unsolved problem in structural biology for 50 years: given a protein's amino acid sequence, predict its three-dimensional structure. The structure determines function; the function determines mechanism; the mechanism determines whether a drug can intervene. The computational solution arrived at the CASP13 competition in December 2018 when DeepMind's AlphaFold v1 outperformed every other method. At CASP14 in November 2020, AlphaFold 2 achieved a median accuracy of less than one angstrom — essentially solving the problem to within the precision limits of experimental crystallography. The July 2021 Nature paper by John Jumper, Demis Hassabis, and 36 co-authors described the architecture; in 2022 DeepMind released predictions for the entire human proteome (20,000 proteins) and then for 200 million proteins across all sequenced organisms — freely available. AlphaFold 3 (May 2024) extended prediction to protein-ligand and protein-nucleic acid complexes, enabling drug-binding-site prediction. In October 2024, Jumper and Hassabis received the Nobel Prize in Chemistry for the work. For medical scientists, AlphaFold eliminated what had previously been a years-long bottleneck: you can now download a high-confidence structure prediction in seconds for a target that would have required 5 years of X-ray crystallography or cryo-EM work. The effect is not job elimination — it is a profound acceleration of the hypothesis-generation stage of drug discovery.

    Effect on the work

    DeepMind reports over 3 million researchers across 190 countries have used AlphaFold since 2022. AI drug discovery companies built on AlphaFold and related tools — Recursion Pharmaceuticals (IPO 2021), Insitro, Schrödinger, Insilico Medicine — collectively raised over $3B in venture capital between 2020 and 2024, creating a new category of computational medical scientist roles at the intersection of ML and structural biology.

    Work toolChanging equipment
  • mRNA platform + AI-accelerated drug discovery + LLM research tools

    Three overlapping technological platforms reshaped what a medical scientist does in the 2020s. First: the mRNA platform validated at scale by COVID vaccines. Katalin Karikó and Drew Weissman's 2005 discovery that chemically modified nucleoside bases could suppress the immune response to synthetic mRNA — work for which they received the 2023 Nobel Prize in Medicine — had languished for a decade before Moderna and BioNTech used it to design vaccine candidates within 48 hours of the SARS-CoV-2 sequence being published (January 11, 2020). The Moderna vaccine received FDA Emergency Use Authorization on December 18, 2020; BioNTech-Pfizer on December 11, 2020 — each under a year from sequence to authorization, shattering the previous record. Over 13 billion COVID-19 doses were administered globally. The platform is now in clinical trials for cancer, influenza, HIV, and RSV vaccines. Second: AI-accelerated drug discovery (Recursion, Insitro, Schrödinger) applies machine learning to high-content screening, ADMET prediction, and hit-to-lead optimization — compressing steps that previously took 3-5 years into months. Third: LLM-based research tools (Elicit, Consensus, Scite, Semantic Scholar) are reshaping the literature-review and hypothesis-generation phases of the job. A medical scientist can now run a structured systematic literature review in hours that previously took months.

    Effect on the work

    The mRNA validation created a commercial-scale platform requiring an entirely new workforce: mRNA design scientists, lipid nanoparticle formulation chemists, and process development researchers. Moderna grew from ~800 employees (2019) to over 5,000 (2022). BioNTech grew from ~1,300 to over 4,500 in the same period. LLM tools are restructuring time allocation within the role but have not reduced headcount — they have shifted time from rote literature processing toward hypothesis testing and experimental design.

    AI audit toolsPattern detection
Projection cone · present → 2034

What credible sources project

Scrub the slider past now to anchor each scenario on the scrubber. The spread is the range of futures credible sources project for this role.

Employment outlook
Projected change in the number of people doing this work.
WEF Future of Jobs Report 2025
2030
+12%
WEF surveys across 1,000+ employers covering 14 million workers globally. Life and biological scientists — the category that encompasses Medical Scientists — are listed among the fastest-growing roles through 2030. The +12% approximation represents the WEF's biomedical research growth signal; the primary drivers are AI-enabled drug discovery (increasing research productivity and thus demand for scientists to run more experiments), aging population demographics across all OECD countries, and sustained public-sector investment in pandemic preparedness and cancer research. The WEF explicitly models AI as complementary to biomedical researchers rather than substitutive in the near term.
O*NET / BLS Occupational Outlook Handbook 2024
2034
+11%
O*NET synthesizes the BLS OOH 2024-34 projections with the occupational classification. The OOH page (which returned HTTP 403 in this curation pass) is reported by O*NET as projecting "much faster than average" growth at approximately 11% — a commonly cited figure in secondary sources describing the 2024-34 cycle for this occupation. The 11% and 8.7% figures reflect different rounding or vintage of the same BLS data; both are directionally consistent. Annual openings of ~9,600 include new positions plus replacement need from retirements.
BLS National Employment Matrix 2024-34
2034
+8.7%
BLS Employment Projections 2024-34 cycle. Baseline employment 165,300 (2024); projected 179,600 (2034); absolute change +14,300 positions. Annual job openings: approximately 9,600 per year. BLS describes the growth as "much faster than average" (defined as 7% or higher). Primary demand drivers: aging population increasing the burden of chronic disease requiring research-backed treatment; biopharmaceutical R&D expansion in mRNA therapeutics, gene therapy, and oncology immunotherapy; federal biomedical research infrastructure maintained at approximately $48B/year NIH budget. The matrix does not explicitly model AI-driven productivity effects.
AI task exposure
Share of the role’s tasks that researchers estimate AI can do. This is a measure of task exposure, not a forecast of jobs lost.
Eloundou et al. — "GPTs are GPTs" (2023)
2028
8%
of tasks
GPT-4 task-by-task LLM exposure labeling on O*NET tasks for life/medical science occupations. Medical Scientists score high on LLM exposure for tasks classified as "information synthesis and writing" — literature review, grant writing, research protocol documentation, regulatory submission writing — but low for core experimental tasks (designing assays, operating instruments, analyzing data from novel experiments requiring judgment). The -8% figure represents the curator-estimated β (exposure with LLM tools) for the writing-and-synthesis component of the role, not total employment displacement. Eloundou et al. explicitly distinguish exposure from displacement; high exposure enables augmentation (faster literature review, better-drafted grants) rather than replacement. The overall β for the full task bundle is moderate; the γ (any exposure) is high because most scientists write.
Frey & Osborne (2013)
2033
5%
of tasks
Gaussian-process classifier on O*NET task features. F&O assigned Medical Scientists a probability of computerization of approximately 0.054 — placing them in the lowest decile of their 702-occupation dataset. The bottleneck factors: high scores on "originality" (generating novel ideas), "scientific knowledge," and "complex problem solving." The -5% figure here is a theoretical lower bound representing the ceiling of the F&O scenario if realized at full strength; actual employment has grown substantially since 2013. F&O explicitly note that occupations requiring "original, creative, and judgment-intensive thinking" are poor candidates for computerization under their 2013 assumptions.
Today, in this role

What's shifting in the work right now

The historical view above shows how this role has moved. This is the present-day detail: which AI tools are picking up which tasks, where the edge still is, and the natural directions this work can grow.

What's changing in your day

Three parts of your work where AI is already doing real lifting, and what stays yours.

AI is sitting alongside you herePredict and analyze protein structures for drug target characterization — using AlphaFold 3 (DeepMind AlphaFold Server or local deployment) to predict three-dimensional structures of disease-relevant proteins and their co-complexes with small molecules, DNA, or RNA

Predict and analyze protein structures for drug target characterization — using AlphaFold 3 (DeepMind AlphaFold Server or local deployment) to predict three-dimensional structures of disease-relevant proteins and their co-complexes with small molecules, DNA, or RNA; interpreting binding pocket geometry, allosteric site accessibility, and conformational flexibility from predicted structures; validating high-confidence predictions against experimental data from PDB before using structural models to guide medicinal chemistry hypotheses.[3],[6],[10]

Where your edge is

AlphaFold 3 is a genuine paradigm shift — predicting structures that previously required months of crystallography or cryo-EM time in under a day. Use it to pre-screen binding site hypotheses before committing to physical structural work, and build fluency with its confidence metrics (pLDDT scores, PAE matrices) so you can identify which predicted regions are reliable enough for medicinal chemistry guidance versus which require experimental validation. The structural biologist who knows how to critically interpret AI-predicted structures — not just run the server — is far more productive than one who either ignores these tools or trusts outputs uncritically.

AI is sitting alongside you hereScreen small molecule compound libraries against drug targets using AI-accelerated virtual screening — using Atomwise AtomNet or Schrödinger FEP+ to screen billions of virtual compounds for binding affinity and selectivity against a disease target

Screen small molecule compound libraries against drug targets using AI-accelerated virtual screening — using Atomwise AtomNet or Schrödinger FEP+ to screen billions of virtual compounds for binding affinity and selectivity against a disease target; applying ADMET (absorption, distribution, metabolism, excretion, toxicity) filters to prioritize compounds with favorable pharmacological properties; selecting a tractable hit list for physical synthesis and biochemical validation; and interpreting false-positive risks given known target family off-target profiles.[13],[14],[10]

Where your edge is

Atomwise and Schrödinger AI can screen billions of compounds in weeks that would require years of physical HTS campaigns — adopt in silico primary screening before committing to physical library synthesis. Your scientific value is concentrated in the interpretation layer: deciding which predicted hits are likely true positives based on target biology (known binding modes, cofactor requirements, allosteric mechanisms), and which ADMET predictions to trust versus challenge with experimental PK/PD data. Build deep knowledge of your target family's structural pharmacology so you can identify when the scoring function is likely to fail (flexible binding sites, covalent mechanisms, allosteric targets) and design the validation tier accordingly.

AI is sitting alongside you hereIdentify clinical trial cohorts and analyze real-world patient data for translational research — using Tempus AI's TIME (Trial Identification and Matching Engine) to query de-identified genomic and clinical data networks to identify patients meeting complex trial inclusion/exclusion criteria

Identify clinical trial cohorts and analyze real-world patient data for translational research — using Tempus AI's TIME (Trial Identification and Matching Engine) to query de-identified genomic and clinical data networks to identify patients meeting complex trial inclusion/exclusion criteria; analyzing genomic profiles alongside clinical outcomes data to identify biomarkers of drug response; and interpreting AI-flagged patient clusters in the context of known disease biology before making enrollment or analysis recommendations.[15],[16],[1]

Tools picking this up
Where your edge is

Tempus AI can compress patient identification for clinical trials from months to weeks, but the translational scientist must verify that AI-identified patients genuinely meet protocol eligibility criteria — particularly complex exclusions based on prior treatment history or comorbidities that are captured inconsistently in structured data. Build a systematic eligibility verification workflow that includes clinical chart review for a randomly sampled validation set before treating AI-identified cohorts as enrollment-ready. The biomarker interpretation step — deciding which genomic co-cluster is mechanistically meaningful versus a statistical artifact — requires disease-specific biological knowledge that Tempus cannot provide.

Where this role is heading

Natural next steps for someone with your foundation: not exits, evolutions.

A direction you could grow

Natural Sciences Managers

Medical scientists who develop program management, team leadership, and cross-functional coordination skills — managing research teams, directing preclinical-to-clinical evidence development, and negotiating industry partnerships — transition naturally into Natural Sciences Manager roles at pharmaceutical companies, biotech organizations, or academic research centers. This path is especially timely as pharma organizations navigate the rapid expansion of AI drug discovery tooling: deciding which AI platforms to adopt, setting review standards for AI-generated molecular candidates, and managing teams of computational biologists alongside wet-lab researchers requires scientific leaders with both research credibility and operational scope. Natural Sciences Managers at pharma and biotech companies earn median $162,760 (BLS 2024), and BLS projects +7% growth through 2034. The PI track record — grant management, publication leadership, trainee mentorship, and vendor/CRO relationships — maps directly to the management skills required.

What you'd add
  • · Research program management: portfolio prioritization across multiple research threads, CRO and vendor governance, milestone-based go/no-go decision frameworks for preclinical drug candidates
  • · AI tool strategy and governance in R&D: evaluating AI drug discovery platforms (BenchSci, Cradle, Recursion), setting organizational standards for AI-generated scientific claims, and managing compute budgets for ML workloads
  • · Industry regulatory literacy: IND application process, FDA pre-IND meeting strategy, CMC (Chemistry, Manufacturing, and Controls) program oversight for biologics and small molecules
  • · People management in science: hiring scientists, conducting performance reviews, career development coaching in academic vs. industry contexts, managing interdisciplinary team dynamics
  • · Business development skills: licensing deal structures for research collaborations, SBIR/STTR grant management, NIH cooperative agreement and pharma sponsored-research agreement (SRA) structures
What it takesSome new skills to pick up
Share this year
Drops anyone you send it to straight into 2026.
Preview card
Part of Science · see all 27roles →
Different role?

See the same long-arc view for your own profession.

Browse the directory by industry, or search by title or SOC code. New roles ship every few weeks. Every profile cites every claim.

Browse all roles

The data behind this timeline

On record since1898
Latest tracked employment172,340 (US, 2025)
Latest median pay$103,410 (2025)
Outlook+8.7% by 2034 (BLS National Employment Matrix 2024-34)
View all 28 cited data points
YearUS employmentMedian annual paySource
19001,500n/aESTIMATE
194012,000n/aESTIMATE
196560,000$9,500ESTIMATE
199085,000n/aESTIMATE
2000104,000$57,000BLS-OEWS
200360,830$59,210BLS-OEWS
200466,450$61,320BLS-OEWS
200573,670$61,730BLS-OEWS
200678,210$61,680BLS-OEWS
200787,440$64,200BLS-OEWS
200899,750$72,590BLS-OEWS
2009101,760$74,590BLS-OEWS
2010112,000$76,700BLS-OEWS
201195,220$76,130BLS-OEWS
201295,420$76,980BLS-OEWS
2013104,280$79,840BLS-OEWS
2014100,740$79,930BLS-OEWS
2015104,440$82,240BLS-OEWS
2016108,870$80,530BLS-OEWS
2017111,690$82,090BLS-OEWS
2018120,320$84,810BLS-OEWS
2019127,180$88,790BLS-OEWS
2020126,110$91,510BLS-OEWS
2021108,550$95,310BLS-OEWS
2022110,550$99,930BLS-OEWS
2023160,000$99,270BLS-OEWS
2024165,300$100,590BLS-OEWS
2025172,340$103,410BLS-OEWS
Embed this timeline on your site

Free for any site. Paste this where the timeline should appear; it stays interactive, every datapoint stays cited, and it sets no cookies on your page. How embedding works

<iframe src="https://futurehistory.earth/embed/19-1042"
  width="100%" height="430" style="border:0"
  title="Medical Scientists, Except Epidemiologists, a Future History timeline"
  loading="lazy"></iframe>

See all roles in Science