Top 10 AI Prompts and Use Cases and in the Healthcare Industry in Durham

Durham and the wider North Carolina health ecosystem are moving AI from pilots to everyday clinical work: Duke's ambient digital scribing now covers 60–70% of primary‑care visits and can return “two hours” per clinician on busy days, cutting after‑hours “pajama time” and letting providers focus on patients (Duke ambient scribing and AI governance at Duke Medicine).

Across the state, tools for lung‑nodule risk scoring, automated patient messaging, image triage and Duke's Sepsis Watch - linked in reporting on “10 ways North Carolina providers are harnessing AI” - show measurable impacts (Sepsis Watch tied to a 31% drop in sepsis mortality) while health systems build monitoring frameworks before broad deployment (How North Carolina providers are harnessing AI).

Those practical gains make Durham a strategic place to gain applied AI skills for health teams and local tech talent.

Bootcamp	Length	Early Bird Cost	Registration
AI Essentials for Work	15 Weeks	$3,582	Register for AI Essentials for Work (15‑week bootcamp)

“On clinical days, I easily get two hours back.” - Dr. Eric Poon, Duke Medicine

Methodology for choosing the top 10 use cases emphasized three local, practical filters: demonstrated operational effect in Durham health systems, measurable workforce implications, and clear pathways for sustainable adoption.

Preference went to use cases tied to existing hospital infrastructure - especially those that integrate with Epic to streamline messaging, scheduling, and billing (AI Essentials for Work: Epic integration for operational efficiency) - because reductions in admin friction directly improve patient throughput and clinician capacity.

Cases were also weighted for workforce impact using local labor analysis and re‑skilling needs (Job Hunting Bootcamp: AI's impact on Durham healthcare jobs and reskilling strategies), and for sustainability via regional talent channels and university–startup partnerships that grow implementation capacity (Full Stack Web + Mobile Development: growing Durham's AI talent pipeline).

The result: a practical list focused on immediate wins that scale without leaving local staff behind.

Atrium Health Wake Forest Baptist in North Carolina has moved an AI‑enabled nodule pathway into routine care, deploying Optellum's Virtual Nodule Clinic to score incidental CT nodules and classify patients as high, intermediate, or low risk - an algorithm trained on more than 70,000 CT scans that helps radiology and pulmonary teams prioritize biopsies and avoid needless procedures.

Read the Wake Forest Baptist news release on AI and robotics for lung cancer: Wake Forest Baptist news release on AI and robotics for lung cancer.

The program pairs that risk model with robotic bronchoscopy to reach small, hard‑to‑access nodules, speeding diagnosis when earlier detection can be decisive - small tumors treated early can improve five‑year survival from about 20% for late‑stage disease to as high as 90% - and it represents one of the clearest local examples of AI changing clinical triage and biopsy yield in North Carolina.

See the Optellum case study on the Virtual Nodule Clinic at Atrium Health: Optellum Virtual Nodule Clinic case study at Atrium Health.

Tool	Details
Tool	Optellum Virtual Nodule Clinic
Training data	>70,000 CT scans
Risk categories	High / Intermediate / Low
Companion tech	Robotic bronchoscopy (Intuitive)

“The exciting part of this artificial intelligence lung cancer prediction tool is that it enhances our decision making, helping doctors intervene sooner and treat more lung cancers at an earlier stage.” - Dr. Christina Bellinger

OrthoCarolina in Charlotte uses the AI-based digital assistant Medical Brain to follow up with hip and knee replacement patients, providing 24/7 personalized guidance, answering recovery questions, and routing red-flag concerns to clinical triage; a four‑month pilot engaged roughly 200 patients - about 30–60 messages per patient - and cut post‑surgery messages and calls to the clinic by about 70%, freeing staff time while keeping a clinician team reviewing every interaction (OrthoCarolina Medical Brain postoperative follow‑up).

The rollout builds on a formal strategic partnership that brings healthPrecision's Medical Brain platform to OrthoCarolina's network of more than 300 providers at nearly 40 locations, aiming to boost value‑based care by automating routine check‑ins and surfacing emerging risks sooner (OrthoCarolina–Medical Brain partnership announcement).

The takeaway for North Carolina systems: targeted postop AI can cut inbound workload dramatically while maintaining clinician oversight, making it a practical first step for teams seeking measurable operational wins.

Item	Detail
Tool	Medical Brain (healthPrecision)
Pilot patients	~200
Messages per patient	30–60 (over 4 months)
Reduction in clinic messages/calls	~70%
OrthoCarolina network	>300 providers; nearly 40 locations

“For decades, OrthoCarolina has been committed to providing patient‑first comprehensive care across a wide array of orthopedic specialties, and the integration of Medical Brain® into our care continuum will help us to better meet patients' real‑time needs while also accelerating our organizational value‑based care goals.” - Dr. Bruce Cohen

Viz.ai's image‑triage and care‑coordination platform automates detection of suspected large‑vessel occlusion (LVO) on CT and pushes real‑time alerts and mobile image views to stroke teams, helping emergency radiology prioritize patients and shorten transfer and treatment timelines; Novant Health was the first system in the Carolinas to deploy Viz LVO across Novant Presbyterian and Forsyth Medical Centers, giving every stroke specialist smartphone access and streamlining triage (Viz.ai care coordination platform, Novant Health deployment).

Viz.ai's neuro portfolio and published implementation data show substantial door‑to‑treatment improvements in real networks, illustrating how imaging AI plus coordinated workflows can turn minutes into better outcomes for centers that treat hundreds to thousands of stroke patients annually in North Carolina (Viz.ai neurovascular impact and time‑saving data).

For Durham‑area systems, the clear takeaway: image‑first AI that routes critical studies to clinicians' phones is a practical, scalable way to shave precious minutes when every minute of ischemia costs brain cells.

Tool	Primary use	Local deployment	Regulatory notes
Viz LVO	Automated LVO detection + mobile alerts	Novant Presbyterian & Forsyth Medical Centers	Viz.ai platform: 50+ FDA‑cleared algorithms; SDH 510(k) announced Jun 12, 2025

“Time is very critical for the brain and we need to shave off minutes every opportunity we can.” - Dr. Laurie McWilliams, Novant Health

WakeMed in Raleigh uses generative AI to draft patient‑portal replies, route routine requests to non‑physician teams, and partner on streamlined refill workflows, cutting inbound messages by about 12–15 per provider per day and materially easing inbox burden for frontline clinicians (WakeMed patient portal message reduction using AI to draft portal messages).

Drafts are presented for clinician review rather than sent automatically, preserving clinical oversight while making triage faster; that operational mix - technology plus routing and role‑based workflows - mirrors other systems that successfully reduced message volume and clinician time in the inbox (North Carolina health systems harnessing AI for patient communication).

Patient studies show AI‑written replies can feel more detailed and empathetic but that satisfaction falls when authorship is disclosed, so WakeMed's next practical step is pairing drafting efficiency with clear communication policies and triage rules to protect trust while reclaiming clinician time for direct care (patient satisfaction with AI‑generated portal messages and disclosure effects).

“The good news is that we have been successful at engaging our patients to stay in better contact with us, but many of us were not operationally prepared for the significant increase in time that needs to be spent addressing these messages.” - Neal Chawla, MD, CMIO, WakeMed

Wake Forest University School of Medicine developed the electronic Cognitive Health Index, an AI‑driven EHR screening tool that flags patients who may have cognitive impairment and who could benefit from specialized programs or Alzheimer's‑related treatments, helping clinicians decide who should be prioritized for further testing, referrals, or medication review (NC Health News - 10 Ways North Carolina Providers Harness AI).

By automatically surfacing higher‑risk patients, the index focuses scarce memory‑care resources and medication decisions on those most likely to benefit, reducing missed opportunities for early intervention and targeted outreach (North Carolina Medical Society - 10 Ways Providers Are Harnessing AI).

Tool	Primary use
electronic Cognitive Health Index	Flags patients for cognitive‑impairment programs and Alzheimer's‑related care

At Duke Health, population‑health outreach for breast cancer screening pairs explainable imaging AI with coordinated follow‑up workflows so patients who miss screening or have ambiguous findings are routed to timely care: explainable mammography tools built by Duke engineers aim to “show their work” to radiologists (Duke explainable mammography tools and research), while systemwide use of advanced 3D mammography (digital breast tomosynthesis) improves image clarity and reduces false alarms across Duke's screening sites (Duke 3D mammography screening program).

Those technical gains are matched by governance: Duke's clinical AI program emphasizes narrowly scoped tasks and institutional review processes so outreach nudges, automated image flags, and referral prompts help close screening gaps without removing clinician oversight (Duke Health clinical AI governance and deployment overview).

The practical payoff for North Carolina is concrete - better image triage plus organized follow‑up turns missed appointments into scheduled diagnostic workups, shortening time to biopsy for suspicious findings and increasing the chance of early, treatable cancer.

Metric	Value
Registered clinical algorithms at Duke	54 tools (≈39 using AI)
Screening technology	3D mammography (digital breast tomosynthesis) across Duke sites

“It's very important that AI technology serve the humans. It's very powerful, but it's just math.” - Michael Pencina, PhD

Duke Health's Sepsis Watch is a productionized deep‑learning early‑warning system that continuously scans the EHR - comparing current patients across 86 real‑time variables every five minutes - to flag likely sepsis well before bedside teams might recognize it; the model was trained on tens of thousands of encounters (over 32 million data points) and delivers a median prediction lead time of about five hours, an interval DIHI estimates could save roughly eight lives per month when paired with rapid‑response workflows (Duke Health Sepsis Watch DIHI implementation details).

The program has been tied to large outcome gains in North Carolina hospitals, with published reports of roughly a 27–31% reduction in sepsis mortality after deployment, plus higher SEP‑1 bundle compliance and far fewer false alerts - concrete evidence that tightly integrated AI plus protocolized clinician action can turn early detection into saved lives (HIMSS case study on Duke Health sepsis predictive analytics outcomes).

Metric	Value
Training data	≈42,000+ patient encounters; >32 million data points
Variables monitored	86
Monitoring frequency	Every 5 minutes
Median prediction lead time	~5 hours
Mortality reduction (reported)	~27–31%

“A lot of people develop AI models, but not many are integrating them into clinical practice to improve clinical outcomes. That is a huge differentiator for us at Duke.” - Suresh Balu

Novant Health developed a Behavioral Health Acuity Risk (BHAR) model - built with random forests in collaboration with mental‑health, emergency‑medicine, and psychiatry clinicians - to scan EHR data captured during routine care and assess the likelihood of suicidal behavior; the model can be natively hosted in the electronic record, update in near‑real time, and surface a color‑coded risk score immediately to clinical staff so teams can prioritize high‑risk patients and target scarce behavioral‑health resources quickly (Novant Health BHAR technical overview).

Local reporting calls out this automatic, data‑driven approach as one of the concrete ways North Carolina systems are using AI to flag suicide and other acute risks, making risk visibility part of clinicians' normal workflows rather than a separate pipeline (NC Health News: 10 ways North Carolina providers are harnessing AI), a practical step that shortens the time between risk detection and clinical action.

Item	Detail
Model	Behavioral Health Acuity Risk (BHAR)
Technique	Random forests (machine learning)
Hosting	Native in Electronic Health Record
Update cadence	Near‑real time
Clinical teams	Mental health, emergency medicine, psychiatry
Output	Color‑coded risk score visible in EMR

Duke Health deployed three machine‑learning models trained on thousands of surgical records to tighten operating‑room schedules, and real‑world use across roughly 33,000+ cases made the models about 13% more accurate than human schedulers at predicting case length - reducing schedule overruns, improving throughput, and translating a modest cut in late finishes into tangible savings (about $79,000 in potential overtime reductions over four months in the study).

By surfacing more reliable case‑time estimates to human schedulers - who “run the board” and validate predictions - Duke shortened delays, increased the number of cases that fall within acceptable time windows, and sped patient access to surgery without removing clinician oversight; the work is now implemented across Duke University Hospital and illustrates a practical Durham‑area win: small improvements in prediction accuracy can free OR capacity and lower labor costs while keeping schedulers central to the workflow (Duke Health algorithm improves accuracy of scheduling surgeries - corporate Duke Health release, Duke Medicine overview: Artificial intelligence in health care - promise and pitfalls).

Metric	Value
Cases used in deployment	≈33,000–33,815
Models trained	3 machine‑learning models
Accuracy improvement vs. humans	~13%
Estimated overtime savings	≈$79,000 (over 4 months)
Current status	In use at Duke University Hospital

“One of the most remarkable things about this finding is that we've been able to apply it immediately and connect patients with the surgical care they need more quickly.” - Daniel Buckland, M.D., Ph.D.

UNC Health is piloting an internal generative AI chatbot - branded in reporting as Ava - that runs on Microsoft's Azure OpenAI Service within a secure, governed environment and integrates with Epic to answer UNC‑specific questions, surface concise how‑to guidance from the system's full training and education library, and draft routine patient‑portal replies for clinician review; the pilot began in June 2023 with a small group of clinicians and administrators and is intended for broader rollout after identifying high‑value use cases, which matters because it reduces time spent sifting through hundreds of documents so teammates can spend more time with patients across a network of 15 hospitals, 19 campuses and more than 900 clinics (UNC Health generative AI pilot announcement) - local reporting also notes UNC fed its training libraries into the model so staff get concise, actionable answers instead of dozens of links (MedCity News report on Ava generative AI and clinician burnout), a practical step that directly targets clinician burnout and administrative drag in North Carolina health systems.

Item	Detail
Tool	Ava (internal generative AI chatbot)
Platform	Azure OpenAI Service; Epic generative AI program
Pilot participants	Initial small group of clinicians & admins (≈30 reported)
Network coverage	15 hospitals, 19 campuses, >900 clinics
Primary use	Answer UNC‑specific queries, summarize training docs, draft portal messages for review

“This is just one example of an innovative way to use this technology so that teammates can spend more time with patients and less time in front of a computer.” - Dr. David McSwain

Practical next steps for beginners in Durham: start with hands‑on, workplace‑focused learning so you can translate examples you already see locally - like WakeMed's AI‑drafted portal replies, Duke's Sepsis Watch alerts, and Atrium/Wake Forest's nodule scoring - into on‑the‑job wins (North Carolina providers harnessing AI - 10 examples).

A concrete first move is a short, applied course that teaches prompt writing, safe tool use, and how to map AI to admin workflows; the 15‑week AI Essentials for Work bootcamp covers those exact skills and is designed to help people in non‑technical roles prototype prompt‑driven fixes for inbox overload, follow‑up outreach, and triage tasks that Durham systems are automating now (AI Essentials for Work syllabus and registration - 15‑week bootcamp).

Pair training with a local project - volunteer to pilot a messaging workflow or shadow an informatics team - so the first portfolio item you build is a measurable operational improvement (fewer messages, faster follow‑ups, or clearer triage), not just a demo; that practical focus is what turns short courses into visible value for North Carolina health employers.

Bootcamp	Length	Early Bird Cost	Registration
AI Essentials for Work	15 Weeks	$3,582	Register for AI Essentials for Work - 15‑week bootcamp