Overarching Legislation:FD&C Act (1938) what counts as a drug or device · HIPAA (1996) privacy baseline · HITECH (2009) $30B+ to adopt EHRs · ACA (2010) Medicaid expansion, marketplaces, CMMI, ACOs, Sunshine Act · MACRA (2015) quality-based Medicare pay · 21st Century Cures (2016) information blocking, APIs, CDS carve-out, RWE framework. Most of the agencies below sit inside HHS; the rest are the FTC, DEA, Education, the courts and the states.
FTC · data handling rules for consumer apps that fall outside of HIPAA: deceptive claims (UDAP), breach notice (HBNR), COPPA for under-13s
FDA · safety, efficacy, marketing rules. CDRH (devices, including SaMD): is it a device? Intended use and claims decide, not the hardware. General wellness → not regulated (see the 2026 update below). CDS → exempt only if the clinician can check the basis. SaMD → Class I / II / III, cleared via 510(k), De Novo or PMA; Breakthrough Devices speeds review; AI devices need GMLP and a PCCP; connected devices, cybersecurity (524B). CDER / CBER (drugs, biologics): IND → trials → NDA / BLA; a drug packaged with an app is a combination product; DTC drug ads are policed by OPDP. Investigational use of anything needs an IND or IDE plus an IRB. RWD/RWE from devices and apps increasingly count as evidence. TEMPO: use inside CMS's ACCESS Model before clearance.
OCR · HIPAA for covered entities (providers, plans, clearinghouses) and their business associates (BAAs): all BA's (like digital health companies) need to handle data according to HIPAA requirements and are liable for information breaches · ACA Section 1557: no discriminatory algorithms or decision tools
Department of Education · FERPA governs student records, including a lot of school health data, instead of HIPAA
States & AGs · privacy laws (CCPA/CPRA, CMIA, MHMDA); AI laws (CA, CO, IL, NY); Attorneys General investigate and sue
ONC / ASTP · certifies EHRs (CEHRT: USCDI, FHIRAPIs), enforces information blocking rules to allow data exchange, runs TEFCA
Public health & local government · CDC and state agencies run registries (IIS, PDMP) and surveillance; local budgets fund school health
Courts · product liability and malpractice apply with no new law needed (see the Character.AI and OpenAI suits)
OIG / DOJ · AKS ('free' software needs a safe harbor), Stark, FCA · Sunshine Act: payments to doctors go public
Buyer gates · not law, but required to sell: SOC 2 / HITRUST, a signed BAA, EHR integration, an IRB for any pilot with patients
States · licensure & scope of practice (only humans), CPOM → MSO + PC, IMLC · DEA controlled substances Rx · SAMHSA42 CFR Part 2
CMMI · tests new payment models: ACCESS (tech-enabled chronic care; FDA's TEMPO rides along), ASPIRE & InCK (Medicaid kids), ACOs
EU · GDPR, the MDR and the AI Act apply the moment you have a European user
New in 2026 · FDA's January 6, 2026 guidance.General wellness: a non-invasive wearable can now estimate blood pressure, blood glucose, SpO2 or HRV and stay outside FDA oversight, as long as the claims stay wellness-only (sleep, activity, stress): no disease names, no clinical thresholds, no 'medical grade' claims, no ongoing disease monitoring; a 'consider seeing a clinician' nudge is fine. CDS: software may now propose a single treatment plan rather than a list of options, as long as the clinician reviews, revises and finalizes it and can see the basis; analyzing images or signals is still a device.
Patient & Family Generates the data: wearables, apps, portal visits. Pays premiums, taxes, copays and sometimes a D2C subscription. About half of US kids are on Medicaid or CHIP.
Pharma & MedTech Drug, biologic and device makers. To the DH company, they are a customer, a data provider, or a distribution channel (the app ships with the device/pharma product). Sometimes the Pharma or MedTech company is the DH company.
Digital Health Company App, device, SaMD, DTx, SaaS. Or, a pharma or device company's digital arm. Sells D2C, by subscription (SaaS, PMPM), or bills insurance itself, which requires licensed clinicians (MSO + PC). HIPAA applies to DH company through a BAA.
Providers: Clinicians & Health Systems Generally, only licensed humans can deliver care. The EHR records it in SNOMED CT, LOINC, RxNorm; bills it in ICD-10-CM + CPT/HCPCS. Share records via HIE, TEFCA/QHINs; report to IIS, PDMP.
Health Insurance Medicare (Parts A, B, C = MA, D). Medicaid & CHIP, mostly via MCOs (EPSDT guarantees kids' benefits). Commercial & employer plans (ESI, self-insured via a TPA), PBMs. Funded by premiums and taxes; CMS sets the fee schedule.
Schools, CBOs & Public Agencies The other sector, especially for kids: school-based health (bills Medicaid), CBOs, local programs, public health. School records: covered under FERPA, not HIPAA. Pays for prevention whose savings aren't seen in the health sector.
Claims: X12 837 in, 835 remittance back Prior auth · quality reporting (eCQMs)
Patient-generated data: wearables, apps, RPM/RTM streams HIPAA and Consumer Protection laws apply to data handling
Ad hoc data sharing: screenings, referrals, attendance, program outcomes (data use agreements, FERPA consent)
Visits and the patient portal: notes and results released after they're signed/finalized Patient-access FHIRAPI: the EHR must hand the record to any app the patient picks (e.g., Apple Health)
Patient Access API (FHIR): the plan must hand claims and clinical data to whatever app the member picks
Buys companion apps, patient support programs, RWE and clinical trial recruitment
Capitated revenue: subscription (SaaS, PMPM) paid by the health system
Medicaid pays school-based health centers (SBHCs) and EPSDT services
Direct to consumer (D2C): subscriptions, app stores, HSA/FSA dollars
Capitated revenue from plans & employers (PMPM / PEPM), or billable to insurance: the company bills the payer itself for eligible supplies or health services
Other payers: grants & contracts from schools, CBOs, public agencies The wrong-pocket problem: who pays isn't who saves
Premiums (plus the taxes that fund Medicare and Medicaid); cost sharing (copays, deductibles) when care is used
Questions, comments, or feedback? Email aydinz@stanford.edu. Many thanks to Meghana Nerurkar for inspiring the creation of this diagram. Learn more about our Biodesign for Digital Health course here.