SCAIO Learn · Primer 06

Healthcare AI in South Carolina.

What clinical AI deployment looks like in SC today — for clinicians, hospital boards, and patient advocates.

SCAIO · scaio.org
The picture in South Carolina

SC is among the most active states in the country for applied clinical AI.

The combination of MUSC's Center for AI, the joint Clemson–MUSC AI Hub for Healthcare and Life Sciences, Hollings Cancer Center (an NCI-designated comprehensive cancer center), and the deployment capacity of Prisma Health and the broader hospital network makes South Carolina an unusually substantive applied-AI environment for healthcare.

The question for SC is not whether clinical AI will arrive. It already has. The question is how it gets deployed.

What's already in use in SC

Four categories of clinical AI currently deployed.

01 · Ambient documentation

DAX Copilot at MUSC

Voice-driven AI that listens to a clinical encounter and produces draft notes for the EHR. Among the highest-volume clinical-AI deployments at MUSC.

02 · Diagnostic imaging

AI-assisted radiology and pathology

Image-analysis models that flag potential findings for clinician review. MUSC's Pap-smear AI work has drawn national attention.

03 · Operational AI

Scheduling, billing, prior-auth

Back-office AI for appointment scheduling, claims processing, prior authorization, and revenue-cycle management. The most invisible category to patients.

04 · Predictive risk

Sepsis, readmission, deterioration

Models trained on patient data that flag patients at elevated risk for specific events. Increasingly common in academic medical centers.

What the state has legislated

S.443 — Health Claims and AI.

South Carolina's S.443 establishes guardrails around AI use in insurance prior-authorization decisions — specifically, the requirement that AI tools used to deny coverage be subject to physician review, and that determinations not rest on AI output alone.

The bill addresses one of the clinical-AI use cases that creates the clearest patient-impact risk. It does not address the broader landscape of clinical-AI deployment — that remains a question of institutional governance, vendor contracting, and clinical practice.

The two-tiered risk

A clinical-AI deployment gap that needs naming.

High-resource end

MUSC, Prisma, major systems

Have the resources, governance structures, and clinical-AI expertise to deploy carefully. Active AI governance committees, evaluation infrastructure, and informed clinical leadership.

Rural-hospital end

Smaller systems and critical-access

Deployment decisions often made by vendors and consultants rather than clinicians and patients. Smaller security budgets, less governance bandwidth, more vendor dependency.

The gap between these two tiers is where a thoughtful statewide approach could be most valuable.

South Carolina has the institutional capacity to lead nationally on responsible clinical AI deployment. The question is whether the leadership at MUSC reaches the smaller hospitals fast enough to matter."

A working frame for the SC clinical-AI conversation
Three questions for any clinical-AI deployment

Questions clinicians and hospital boards should ask.

Q1

What evidence shows this tool produces the outcomes the vendor claims, in patients like ours?

Independent peer-reviewed evidence on a comparable population is the floor. Vendor-supplied "case studies" are not evidence.

Q2

How does the tool perform across demographic subgroups?

AI tools trained on non-representative data can perform meaningfully worse on under-represented populations. The bias-audit question is now a standard part of clinical-AI evaluation.

Q3

What happens when the tool is wrong, and who is accountable?

Every clinical-AI tool has failure modes. The institution that deploys it owns those failure modes — not the vendor. The governance question deserves explicit attention.

What would help, statewide

A voluntary clinical-AI deployment standard for SC.

SCAIO's flagship report (Chapter 8, Recommendation 8) proposes a voluntary statewide standard for clinical-AI deployment, jointly developed by MUSC and the SC Hospital Association.

The standard would not be a mandate. It would be a published reference — covering evidence requirements, bias auditing, patient-disclosure practices, governance structures, and a deployment template smaller systems can adopt without rebuilding it themselves.

The work is supported by national reference frameworks (Coalition for Health AI / CHAI), peer-state precedents (Texas, California), and SC's existing institutional alignment.

Where to go for help in South Carolina

SC-specific resources for healthcare AI work.

Academic

MUSC Center for AI

Clinical-AI research, deployment governance, and applied research portfolio at the state's largest academic medical center.

Academic

Clemson–MUSC AI Hub for Healthcare and Life Sciences

Joint hub combining Clemson's computational research with MUSC's clinical environment. Strong applied-research portfolio.

Member organization

SC Hospital Association

Statewide hospital member organization. Natural co-convener for a SC-specific clinical-AI deployment standard.

Policy

SC Department of Insurance

S.443 oversight authority. Where prior-authorization questions and disputes route.

For patients

Three things worth knowing.

SCAIO Learn

Public-interest AI research for South Carolina.

SCAIO tracks AI's impact on South Carolina healthcare and other public-interest sectors. Browse more primers and the flagship report at scaio.org.

scaio.org · jimmy@scaio.org

1 / 11 Exit