In 2008, a 22-year-old Wharton graduate named Nat Turner watched a family member navigate a cancer diagnosis and discovered something that would change the course of his career. The oncology research system — supposedly the most data-driven corner of medicine — was studying only about four percent of cancer patients’ records in clinical trials. The other 96 percent of patient data sat locked inside incompatible electronic health record systems, handwritten notes, and siloed databases across thousands of community oncology clinics. Treatments were being developed in a near-vacuum of real-world evidence. Turner, who had already built and sold an advertising technology company to Google for over $70 million, realized that the same data infrastructure techniques transforming digital advertising could be applied to cancer research. In 2012, he and his longtime collaborator Zach Weinberg founded Flatiron Health — a company that would aggregate, structure, and analyze oncology data at a scale the medical world had never seen. Six years later, pharmaceutical giant Roche acquired Flatiron for $1.9 billion, validating the thesis that real-world clinical data, properly organized, could accelerate how the world fights cancer. Turner’s journey from teenage reptile breeder to health tech billionaire is one of the most unusual entrepreneurial arcs in modern technology.
Early Life and the Entrepreneurial Instinct
Nathaniel Turner grew up in Chevy Chase, Maryland, in a family that encouraged curiosity but had no particular connection to technology or entrepreneurship. His first ventures were decidedly analog. As a teenager, he bred and sold reptiles out of his parents’ garage — a business that taught him supply chain management, customer acquisition, and the fundamental mechanics of buy low, sell high. He also ran a trading card business and, crucially, began building websites to promote these ventures. The website work became a business of its own: a small web design shop that gave Turner his first exposure to the economics of software.
Turner enrolled at the Wharton School at the University of Pennsylvania, where he concentrated in Entrepreneurship and Marketing, graduating cum laude with a Bachelor of Science in Economics. At Wharton, he met Zach Weinberg — a partnership that would define both of their careers. The two recognized each other as complementary thinkers: Turner was the product visionary and external-facing leader; Weinberg was the operational and technical architect. Together, they would found two companies worth a combined $2 billion before either of them turned 35.
During their time at Penn, Turner and Weinberg immersed themselves in the early programmatic advertising ecosystem. Display advertising in the mid-2000s was a fragmented mess — advertisers had to negotiate directly with individual publishers, placement was manual, and there was no systematic way to optimize which ads appeared where. Turner and Weinberg saw that the entire system could be automated using real-time bidding and data-driven decision-making. This insight led to their first company.
Invite Media: The Proving Ground
In 2007, Turner and Weinberg co-founded Invite Media, which built one of the industry’s first demand-side platforms (DSPs) for programmatic advertising. The platform allowed advertisers to bid on display ad inventory in real-time, using data signals to determine which impressions were worth buying and at what price. It was, at its core, a data pipeline problem: ingest massive volumes of impression-level data, apply algorithmic optimization in milliseconds, and execute transactions at scale.
Invite Media’s technology was ahead of its time. The company built infrastructure for processing billions of ad impressions daily, applying machine learning models to predict click-through and conversion rates, and executing real-time auctions across multiple ad exchanges simultaneously. The platform’s architecture — designed for high throughput, low latency, and data-driven decision-making — would prove directly relevant to Turner’s later work in health technology.
Google acquired Invite Media in June 2010 for a reported $81 million. Turner was 24 years old. The acquisition became a key component of Google’s DoubleClick Bid Manager (now Display & Video 360), which remains one of the dominant programmatic advertising platforms in the world. Turner and Weinberg stayed at Google briefly, but the scale and bureaucracy of a large corporation did not suit them. They left to find their next problem — and this time, they wanted something with more meaningful impact than banner ad optimization.
The Genesis of Flatiron Health
The personal experience that catalyzed Flatiron Health was not a sudden revelation but a slow-building frustration. When Turner’s family member was diagnosed with cancer, Turner expected to find a sophisticated, data-rich research ecosystem — the kind of infrastructure that technology companies had built for far less consequential problems. Instead, he found fragmentation and opacity. Clinical trial data covered a tiny fraction of patients. Community oncology practices — where the majority of cancer patients in the United States receive treatment — used electronic health record systems that were not designed for research. Patient records contained unstructured text, scanned documents, and handwritten notes that no algorithm could easily parse.
Turner and Weinberg spent months researching the oncology data landscape before writing a single line of code. They interviewed oncologists, data scientists, pharmaceutical researchers, and FDA officials. They mapped the entire workflow of how cancer data moved (or failed to move) from a patient’s clinic visit to a published research finding. What they found was a system where the data existed in abundance — every cancer patient generates enormous volumes of clinical data — but the infrastructure to aggregate, structure, and analyze that data simply did not exist.
In 2012, Turner and Weinberg founded Flatiron Health in New York City’s Flatiron District (the name was a deliberate nod to the neighborhood). The company’s thesis was direct: build the technology platform that connects oncology data across the fragmented American healthcare system, and use that connected data to accelerate cancer research. Their initial funding of $8 million came from Google Ventures (now GV), First Round Capital, and LabCorp — a strategic mix of technology venture capital and healthcare industry expertise.
Building the Oncology Data Platform
The Technical Architecture
Flatiron Health’s technical challenge was fundamentally different from anything Turner had built at Invite Media, yet the architectural principles transferred. At Invite Media, the problem was processing billions of ad impressions in real-time to make bidding decisions. At Flatiron, the problem was ingesting heterogeneous clinical data from hundreds of oncology practices, normalizing it into a structured format suitable for research, and doing so with the accuracy that medical decision-making demands.
The company’s platform consisted of several interconnected systems. OncoEMR, an oncology-specific electronic health record system acquired when Flatiron purchased Altos Solutions in 2014, served as the data collection layer. OncoEMR was the first web-based EMR built specifically for oncology practices, and it captured structured clinical data at the point of care — treatment regimens, lab results, staging information, genomic testing results, and clinical outcomes. The system included over 3,000 NCCN Order Templates and AJCC staging content, ensuring that data was captured in standardized formats.
But the real innovation was Flatiron’s data curation pipeline — a hybrid system combining machine learning with human clinical expertise. Automated algorithms extracted data from medical records, including unstructured clinical notes, pathology reports, and imaging results. A team of trained oncology nurses and clinical data specialists then reviewed and validated the algorithmically extracted data. The discrepancies between machine extraction and human review were fed back into the models, creating a continuous improvement loop that steadily increased accuracy over time.
# Simplified clinical data pipeline architecture
# Illustrating Flatiron's hybrid ML + human curation approach
class OncologyDataPipeline:
"""
Core pipeline for transforming raw clinical records into
research-grade structured oncology data.
Flatiron's innovation: combining NLP-based extraction with
human clinical review to achieve research-grade accuracy.
"""
def __init__(self, config):
self.nlp_extractor = ClinicalNLPEngine(config.models)
self.validator_pool = ClinicalReviewTeam(config.nurses)
self.feedback_loop = ModelRetrainingPipeline()
def ingest_patient_record(self, raw_record):
"""Process a single patient record from OncoEMR."""
# Step 1: Parse structured fields (labs, meds, staging)
structured = self.parse_structured_fields(raw_record)
# Step 2: NLP extraction from unstructured clinical notes
extracted = self.nlp_extractor.process(
notes=raw_record.clinical_notes,
pathology=raw_record.pathology_reports,
imaging=raw_record.radiology_reports
)
# Step 3: Merge structured + extracted data
merged = self.merge_data_sources(structured, extracted)
# Step 4: Route to human review based on confidence score
if extracted.confidence < REVIEW_THRESHOLD:
validated = self.validator_pool.review(merged)
# Feed discrepancies back into model training
self.feedback_loop.log_discrepancy(
machine=extracted,
human=validated,
record_id=raw_record.id
)
return validated
return merged
def build_research_cohort(self, criteria):
"""
Query the curated database to build patient cohorts
for real-world evidence studies.
Criteria example: advanced NSCLC patients who received
immunotherapy as second-line treatment after 2018.
"""
cohort = self.query_engine.filter(
tumor_type=criteria.tumor_type,
stage=criteria.stage_range,
treatment_line=criteria.line_of_therapy,
drug_class=criteria.drug_class,
date_range=criteria.treatment_period
)
return DeidentifiedCohort(
patients=cohort,
variables=criteria.requested_variables,
linkage=criteria.allow_genomic_linkage
)This hybrid approach — machines for speed, humans for accuracy, feedback loops for improvement — was central to Flatiron’s competitive advantage. Pure automation could not achieve the accuracy required for medical research. Pure manual curation could not scale to millions of patient records. The combination of both, with systematic learning from disagreements, produced a dataset that the FDA would eventually trust for regulatory decision-making.
The OncoCloud Ecosystem
Around OncoEMR, Flatiron built a suite of tools collectively called OncoCloud. OncoBilling handled the complex billing workflows specific to oncology practices (where a single chemotherapy session can involve dozens of separate charges). OncoAnalytics provided practice-level dashboards showing treatment patterns, outcomes, and operational metrics. OncoTrials helped match patients to relevant clinical trials — addressing one of the most persistent problems in oncology research: the fact that fewer than five percent of adult cancer patients participate in clinical trials, largely because physicians and patients are unaware of available studies.
By the time of the Roche acquisition, Flatiron’s network encompassed over 265 community cancer clinics, six major academic research centers, and partnerships with 14 of the top 15 therapeutic oncology companies. The de-identified patient database contained records from millions of cancer patients across more than 22 tumor types — a dataset unmatched in oncology research for its breadth, depth, and longitudinal coverage. This data platform enables machine learning approaches to identify treatment patterns that would be invisible in traditional clinical trial data.
The Data Model That Changed Oncology Research
What made Flatiron’s data uniquely valuable was not just its scale but its structure. The company developed an oncology-specific data model that captured the full complexity of cancer treatment in a format suitable for computational analysis. Traditional clinical databases treated cancer as a series of isolated events — a diagnosis here, a prescription there, a lab result somewhere else. Flatiron’s model captured the entire patient journey as a connected timeline, linking diagnoses to treatments to outcomes to genomic profiles.
-- Simplified oncology data model
-- Illustrating how Flatiron structured cancer patient journeys
-- for real-world evidence generation
-- Core patient timeline: every cancer journey as a connected graph
CREATE TABLE patient_journey (
patient_id UUID PRIMARY KEY,
birth_year INT,
sex VARCHAR(10),
race_ethnicity VARCHAR(50),
practice_type VARCHAR(20), -- 'community' or 'academic'
region VARCHAR(30),
first_activity_date DATE,
last_activity_date DATE,
vital_status VARCHAR(20), -- 'alive', 'deceased', 'lost'
death_date DATE
);
-- Diagnosis-level data: a patient can have multiple cancer diagnoses
CREATE TABLE diagnosis (
diagnosis_id UUID PRIMARY KEY,
patient_id UUID REFERENCES patient_journey,
tumor_type VARCHAR(50), -- 'NSCLC', 'breast', 'CRC', etc.
histology VARCHAR(100),
stage_at_diagnosis VARCHAR(10), -- AJCC staging: I, II, III, IV
diagnosis_date DATE,
-- Biomarker panel at diagnosis
egfr_status VARCHAR(20),
alk_status VARCHAR(20),
pdl1_expression DECIMAL(5,2),
msi_status VARCHAR(20),
tmb_score DECIMAL(8,2)
);
-- Treatment episodes: linked to diagnosis, ordered by line of therapy
CREATE TABLE treatment_episode (
episode_id UUID PRIMARY KEY,
diagnosis_id UUID REFERENCES diagnosis,
line_of_therapy INT, -- 1st line, 2nd line, etc.
regimen_name VARCHAR(200), -- e.g. 'pembrolizumab + carboplatin'
drug_class VARCHAR(100), -- 'immunotherapy', 'chemotherapy'
start_date DATE,
end_date DATE,
discontinuation VARCHAR(50), -- 'progression', 'toxicity', 'completed'
best_response VARCHAR(30) -- 'CR', 'PR', 'SD', 'PD' (RECIST)
);
-- This structure enables queries like:
-- "What is the real-world progression-free survival for 2nd-line
-- immunotherapy in PD-L1-high advanced NSCLC patients treated
-- in community oncology settings?"
--
-- That question, answerable in seconds with this data model,
-- would take years and millions of dollars in a traditional
-- clinical trial framework.This data model — connecting patient demographics, molecular profiles, treatment decisions, and clinical outcomes in a single queryable structure — enabled a new kind of medical research. Pharmaceutical companies could study how their drugs performed in real-world populations that differed from carefully selected clinical trial participants. The FDA could evaluate post-market safety signals across millions of patients rather than thousands. Oncologists could benchmark their treatment patterns against national data, using tools similar to agile methodologies adapted for clinical workflows. And researchers could generate hypotheses about treatment sequencing, combination therapies, and biomarker-driven patient selection at a pace that traditional clinical research could never match.
The Roche Acquisition and FDA Partnership
Roche, the Swiss pharmaceutical giant and the world’s largest oncology drug maker, had been an investor in Flatiron since 2014. In February 2018, Roche acquired the remaining shares of Flatiron Health for approximately $1.9 billion, making it one of the largest health tech acquisitions in history at that time. The strategic logic was compelling: Roche spent billions annually developing cancer drugs, and Flatiron’s data platform could make that development faster, cheaper, and more targeted.
Critically, Flatiron maintained operational independence after the acquisition — a condition Turner and Weinberg insisted upon. The company continued to work with all major pharmaceutical companies, not just Roche, and its data platform remained available for industry-wide research. This independence was essential for Flatiron’s credibility as a neutral data partner and for maintaining the trust of the oncology clinics that contributed patient data to the network.
Perhaps the most significant validation of Flatiron’s approach came from the U.S. Food and Drug Administration. Beginning in 2019, the FDA entered into a multi-year research collaboration with Flatiron Health to explore how real-world data could support regulatory decision-making. The collaboration involved analyzing Flatiron’s datasets to assess whether real-world evidence could supplement or, in some cases, replace traditional clinical trial data for specific regulatory questions. This was a paradigm shift: the FDA was acknowledging that data from routine clinical care, properly curated, could contribute to drug approval and safety monitoring.
By 2024, Flatiron had surpassed 1,000 peer-reviewed research publications using data from its de-identified patient database. The platform’s data configurations encompassed more than five million patient journeys across four of the largest oncology markets in the world and more than 22 tumor types. The scale and quality of this dataset made it the gold standard for real-world evidence in oncology — a remarkable achievement for a company that did not exist a decade earlier.
Philosophy and Leadership Approach
From Ad Tech to Health Tech: The Transfer of Principles
Turner’s transition from advertising technology to healthcare might seem like a radical pivot, but the underlying principles were consistent. Both domains involve massive volumes of heterogeneous data, real-time decision-making, and the challenge of extracting signal from noise. The ad tech stack Turner built at Invite Media — ingesting billions of data points, applying predictive models, and executing decisions at scale — was conceptually similar to the clinical data pipeline Flatiron required.
What differed was the tolerance for error. In advertising, a misclassified impression costs a fraction of a cent. In oncology, inaccurate data can influence treatment decisions that affect patients’ lives. This difference drove Flatiron’s hybrid curation model and its obsessive focus on data quality — an approach that resembles how performance optimization in software requires both automated tooling and human judgment to identify meaningful bottlenecks.
Turner has spoken publicly about the importance of domain immersion. When founding Flatiron, he and Weinberg did not assume their technology expertise was sufficient — they spent months embedded in oncology clinics, learning the workflows, the terminology, the regulatory constraints, and the human dynamics of cancer care. This willingness to become genuine domain experts, rather than parachuting in with generic technology solutions, set Flatiron apart from dozens of other startups that attempted to “disrupt” healthcare with Silicon Valley hubris.
Building for Impact, Not Just Scale
Turner’s entrepreneurial philosophy prioritizes impact over growth metrics. While Invite Media was a profitable exit, Turner has described it as ultimately unsatisfying — optimizing which banner ads people saw was not the kind of problem worth dedicating a career to. Flatiron, by contrast, was built around a mission that mattered: using data to improve cancer treatment and accelerate the development of new therapies. This mission-driven approach influenced everything from recruiting (attracting top engineers who wanted their work to have medical significance) to product decisions (prioritizing data accuracy over speed-to-market).
The emphasis on mission also shaped Flatiron’s organizational culture. Turner built a company where engineers sat next to oncologists, where product decisions were informed by clinical reality rather than abstract market analysis, and where the definition of success was measured in research outcomes as much as revenue. This interdisciplinary approach — embedding domain experts deeply within the technology team — is a model that has been adopted by subsequent health tech startups and that parallels how modern development frameworks integrate design thinking directly into the engineering process.
Beyond Flatiron: Ongoing Impact
After the Roche acquisition, Turner transitioned away from day-to-day leadership at Flatiron Health and expanded his activities across multiple domains. He co-founded Operator Partners, a venture fund that invests exclusively with the partners’ own capital — no outside limited partners. The fund writes checks of $250,000 to $1 million alongside lead investors, targeting early-stage companies where Turner and his partners can provide hands-on operational support. His investment portfolio spans more than 38 companies across enterprise software, health technology, and emerging sectors.
Turner also serves as CEO of Collectors, the parent company of Professional Sports Authenticator (PSA), the world’s leading third-party authentication and grading service for collectibles. This role, while seemingly unrelated to health tech, reflects Turner’s consistent interest in building data-driven platforms that bring transparency and standardization to opaque markets — whether those markets involve cancer treatment data or sports memorabilia authentication.
His board memberships and advisory roles span the health tech ecosystem, and he has been recognized by Forbes’ 30 Under 30, Crain’s New York Business 40 Under 40, and the World Economic Forum as a Young Global Leader. More importantly, the model Turner helped create at Flatiron — using real-world clinical data to supplement traditional clinical trials — has become a foundational concept in modern pharmaceutical research and regulatory science. Every major pharma company now has a real-world evidence strategy, and the FDA has issued formal guidance on using real-world data in regulatory decisions. Much of this shift traces back to the infrastructure and credibility that Flatiron Health built under Turner’s leadership. For teams managing the complexity of healthcare technology projects, platforms like Taskee can help coordinate the kind of cross-functional collaboration between engineers, clinicians, and data scientists that Turner championed at Flatiron.
Legacy in Health Technology
Nat Turner’s legacy is not a single invention or a famous algorithm — it is the demonstration that Silicon Valley’s data engineering capabilities could be applied to one of medicine’s most consequential problems with transformative results. Before Flatiron, the oncology research community debated whether real-world data could ever be rigorous enough for regulatory use. After Flatiron, that debate was largely settled. The company proved that clinical data from routine care, when properly structured and curated, could generate evidence that rivaled and sometimes surpassed the insights from randomized controlled trials.
Turner’s career also illustrates a pattern that is increasingly common in modern tech entrepreneurship: the serial founder who applies lessons from one domain to an entirely different one. The data pipeline architecture, the machine learning optimization, the scale-oriented engineering culture — these translated directly from ad tech to health tech. What changed was the purpose, the rigor, and the regulatory context. The ability to recognize that technology built for selling banner ads could be repurposed to fight cancer required both technical sophistication and imaginative leapfrogging — the kind of cross-domain thinking that distinguishes great entrepreneurs from good ones. Building interdisciplinary technology companies with this level of ambition requires robust project coordination, which is where agencies using Toimi can streamline the delivery of complex health tech and digital platform projects.
At 38, Turner has already built two companies worth a combined $2 billion, helped create a new category of healthcare data infrastructure, influenced FDA regulatory policy, and established himself as one of the most effective health tech entrepreneurs of his generation. His story suggests that the most impactful technology companies are not always the ones that invent new science — sometimes, they are the ones that build the infrastructure to connect and organize what already exists, transforming scattered data into coherent knowledge. In oncology, that transformation is literally saving lives.
Key Facts
- Full name: Nathaniel (Nat) Turner
- Born: circa 1986, Chevy Chase, Maryland, United States
- Known for: Co-founding Flatiron Health (oncology data platform acquired by Roche for $1.9B), co-founding Invite Media (ad tech, acquired by Google for $81M)
- Education: B.S. in Economics (Entrepreneurship and Marketing concentrations), The Wharton School, University of Pennsylvania, cum laude
- Key roles: CEO and Co-Founder of Flatiron Health (2012–2018), Co-Founder of Invite Media (2007–2010), CEO of Collectors/PSA, Co-Founder of Operator Partners
- Recognition: Forbes 30 Under 30, Crain’s New York 40 Under 40, World Economic Forum Young Global Leader
- Investments: 38+ companies including Oscar Health, Bell Biosystems, Doctor Evidence, and Tracer Bio
Frequently Asked Questions
Who is Nat Turner and what is Flatiron Health?
Nat Turner is an American technology entrepreneur who co-founded Flatiron Health in 2012 with Zach Weinberg. Flatiron Health is a health technology company that built the leading oncology-specific data platform, connecting electronic health records from hundreds of cancer clinics into a unified, research-grade database. The platform enables pharmaceutical companies, researchers, and regulators to study real-world cancer treatment patterns and outcomes at unprecedented scale. Roche acquired Flatiron Health in 2018 for approximately $1.9 billion. Before Flatiron, Turner co-founded Invite Media, an advertising technology company acquired by Google in 2010.
How does Flatiron Health’s technology work?
Flatiron Health’s platform centers on a hybrid data curation model that combines machine learning with human clinical expertise. Automated natural language processing algorithms extract structured data from unstructured clinical records — physician notes, pathology reports, imaging results, and treatment documentation. Trained oncology specialists then review and validate the machine-extracted data, with discrepancies fed back into the models for continuous improvement. This approach achieves the accuracy required for medical research while scaling to millions of patient records across more than 22 tumor types. The platform’s core product, OncoEMR, is a cloud-based oncology-specific electronic health record system used by over 4,500 cancer care providers.
What impact has Flatiron Health had on cancer research?
Flatiron Health has fundamentally changed how real-world data is used in oncology research. The platform’s de-identified patient database has contributed to more than 1,000 peer-reviewed research publications. The FDA entered into a formal multi-year research collaboration with Flatiron to explore how real-world evidence can support regulatory decision-making — a milestone that helped establish the legitimacy of real-world data in pharmaceutical regulation. Flatiron’s data has been used to study treatment patterns, identify safety signals, benchmark clinical outcomes, and accelerate the design of clinical trials for new cancer therapies.
What is Nat Turner doing after Flatiron Health?
After the Roche acquisition, Turner transitioned from day-to-day leadership at Flatiron Health and diversified his activities. He co-founded Operator Partners, a self-funded venture capital firm that invests $250,000 to $1 million in early-stage companies, with a portfolio spanning more than 38 investments across health tech, enterprise software, and emerging sectors. He also serves as CEO of Collectors, the parent company of PSA (Professional Sports Authenticator), applying data-driven platform principles to the collectibles authentication market. Turner remains active as an angel investor and board member across the health technology ecosystem.
What lessons does Nat Turner’s career offer for tech entrepreneurs?
Turner’s career demonstrates several principles relevant to technology entrepreneurs. First, skills transfer across domains — the data pipeline and machine learning expertise from ad tech translated directly to health tech. Second, domain immersion matters — Turner spent months learning oncology before building technology, rather than assuming generic tech solutions would work. Third, mission-driven companies attract exceptional talent and sustain long-term commitment. Fourth, hybrid approaches that combine automation with human expertise often outperform purely automated solutions, especially in high-stakes domains. Finally, Turner’s experience shows that some of the most valuable technology companies do not invent new science but instead build infrastructure that connects and organizes existing data in transformative ways.
