Joe Roushar

Hidden Pitfalls in EMR Transitions Not all EMR transitions are created equal. Watch out for these common pitfalls that can derail your project: ·        Complex data migration ·        Poor change management planning ·        Hidden costs beyond licensing ·        Workflow mismatches ·        Overbooked vendors with limited capacity The key? Plan early, ask hard questions, and partner with experienced vendors. #HealthcareIT #EMRSelection #CPSI #DigitalHealth #RevenueCycle

20

5 Comments

30,000 users. 34 million chat messages. API keys. Billing credentials. Trade secrets. All leaked because someone trusted an AI aggregator with data they assumed was safe. The OmniGPT breach (February 2025) should be required reading for every enterprise AI leader. Here is what actually happened: A threat actor accessed OmniGPT's application endpoint — likely through flawed session handling or authentication gaps. In one exfiltration, they extracted every conversation users ever had with ChatGPT-4, Claude, Gemini, and Midjourney through the platform. The hacker's own words: "You can find a lot of useful information in the messages — API keys, credentials, billing information." The data was listed on BreachForums for $100. Three hard lessons every enterprise should take from this: 1. Your employees are already sharing sensitive data with AI tools — knowingly or not. Strategy documents. Architecture diagrams. Customer data. HR decisions. It goes into the prompt. It gets stored. It can be stolen. 2. The attack surface of AI platforms is not the model. It is the authentication layer sitting between your users and everything they type. Aggregators, wrappers, and third-party AI tools are the weakest link in your AI security chain. 3. "We use a reputable AI provider" is not a security posture. OmniGPT sat between users and GPT-4, Claude, and Gemini — models from companies with world-class security teams. The breach happened in the middle layer, not the model. Secure-by-design means governing every layer of your AI stack — not just the foundation model. This is not a future risk. It already happened. What are you doing to audit which AI tools your employees are using — and what data is flowing through them? Follow for more on AI Security, Governance, Risk and Sovereign AI — posting weekly. — Source: SecureWorld — OmniGPT Data Breach Exposes 30,000 Users and Millions of Chat Messages (February 2025) #AISecurity #AIGovernance #AIRisk #SovereignAI #CyberSecurity #EnterpriseAI #DataProtection #DPDP

1

Becoming Clinical Interoperability This Post discusses the multiple interconnected terms we use when faced with specifying clinical interoperability scenarios. We will introduce clear definitions and relationships between the core concepts of interoperability in a healthcare context. We are going to illustrate the relationship of the concepts through several interoperability workflow patterns rendered by sequence diagrams. We will focus on the following key workflows: 1. Basic FHIR API Workflow: Patient data retrieval via FHIR API. 2. Medical Device Integration Workflow: Infusion pump sending data to EHR. 3. Cross-Enterprise Document Sharing (XDS) Workflow: Sharing a clinical document between two different healthcare organizations. 4. Laboratory Result Delivery Workflow: From lab device to EHR via LIS and HIE. These diagrams depict the flow of data and the interactions between the clinical interoperability core concepts (Platform, System, Component, Device, API-interface, and Artifact) in various interoperability scenarios. Each diagram highlights the role of standards (like HL7, FHIR, IHE profiles) and the importance of APIs and interfaces in enabling these workflows. A System resides on a Platform. It is composed of several Components. It connects to other Systems and Devices via API-interfaces. It creates, consumes, and manages Artifacts. An Artifact is the data object created by a System, Component, or Device. It is transmitted from one entity to another via an API-interface. It is stored and managed on a Platform. In essence, Platforms provide the foundation, Systems perform the core functions, Components and Devices create and consume data, APIs are the communication channels, and Artifacts are the standardized data packages that flow through these channels. All this, to achieve continuity of care with seamless interoperable, and a medical outcome, that improves the citizen's functional health status. This comprehensive explanation of clinical interoperability concepts and workflows, along with conceptual frameworks, standards compliance regulation, and visual representation of conformance rules (IHE Implementation Guides), provides a holistic view of how healthcare systems achieve interoperability in practice. Enjoy! jvilalta@vico.org https://vicoacademy.com/ #VICOOPENAcademy @VICO OPEN Academy 

10

This video by @InterSystems introduces the OMOP Platform, a new service by InterSystems designed to support researchers by transforming EHR data into the OMOP format. 👉 https://lnkd.in/gAq42nhm The platform features a FHIR-to-OMOP transformation pipeline and an OMOP repository hosted on #InterSystems #IRIS Cloud #SQL.

8

Cross-Industry AI Governance: Lessons from 30 Years of Transformation Having guided transformations across education, healthcare, finance, and government, I've observed how AI governance requirements vary dramatically by industry, yet share fundamental principles. Healthcare: HIPAA compliance intersects with algorithmic accountability. AI decisions affecting patient care require audit trails that trace from data input to clinical recommendation. The governance framework must balance innovation speed with life critical accuracy. Financial Services: Model interpretability isn't optional, it's regulatory. Fair lending laws require explainable AI decisions. Governance must address model bias testing, decision documentation, and regulatory reporting simultaneously. Education: Student privacy meets learning optimization. AI systems that personalize education must protect student data while providing transparent progress tracking. Governance frameworks must address parental consent, data retention, and algorithmic fairness. Government: Public accountability demands extreme transparency. Citizens have the right to understand how AI affects service delivery. Governance must enable innovation while maintaining public trust. The common thread? Stakeholder trust is the ultimate governance metric. Industry-specific regulations provide the compliance floor, but trust requires exceeding minimum requirements. The organizations that thrive will build governance frameworks that treat compliance as the starting point, not the destination. Every industry faces unique AI governance challenges, but the underlying principle remains constant: governance quality determines sustainability.

3

The best CIOs aren’t just digitizing operations — they’re re‑wiring how decisions happen. Every API, workflow, and dashboard is a lever for cultural change. The tech is table stakes; the differentiator is velocity of insight — how fast truth moves through your organization.

11

2 Comments

"Don't use AI at work. Patient privacy." Someone told you this. They lied. HIPAA protects identifiers. Names. Records. Dates. Your ↳Orientation packets ↳Documentation templates  ↳Evidence summaries, hold none of these, so they're free game. Strategic AI use provides you with an avenue to redeem the time redundancy steals from you. The tools exist. It's up to you to master them. [👉🏾 Follow Timi for more | ♻️ Repost if this resonated]

22

13 Comments

𝗪𝗵𝗮𝘁 𝗶𝗳 𝘆𝗼𝘂𝗿 𝗘𝗛𝗥 𝗳𝗶𝗻𝗮𝗹𝗹𝘆 𝘄𝗼𝗿𝗸𝗲𝗱 𝗳𝗼𝗿 𝘆𝗼𝘂 — 𝗻𝗼𝘁 𝗮𝗴𝗮𝗶𝗻𝘀𝘁 𝘆𝗼𝘂? Oracle and Epic are taking very different approaches to AI-powered EHRs. 𝗢𝗿𝗮𝗰𝗹𝗲: built a cloud-native, voice-driven system from scratch. 𝗘𝗽𝗶𝗰: enhancing its legacy platform with in-EHR AI scribes. 𝗛𝗲𝗿𝗲’𝘀 𝘄𝗵𝘆 𝘁𝗵𝗶𝘀 𝗺𝗮𝘁𝘁𝗲𝗿𝘀 ⬇️ → 66% of U.S. physicians now use some form of AI in their workflow (vs. 38% last year). → Ambient scribes cut documentation time by 20–40% and reduce “pajama time” by ~30%. → Early pilots show 40–60% lower burnout symptoms for clinicians. 𝗜’𝗱 𝗹𝗼𝘃𝗲 𝘁𝗼 𝗵𝗲𝗮𝗿 𝘆𝗼𝘂𝗿 𝗽𝗲𝗿𝘀𝗽𝗲𝗰𝘁𝗶𝘃𝗲 ⬇️ → Which features in an AI-powered EHR would change your daily work the most? → Have you tried ambient scribes or voice-driven queries yet? What’s your verdict? #HealthTechInnovation #AIinHealthcare #EHRRevolution #ClinicianExperience #DigitalHealth #AmbientAI #MedicalDocsAI

40

18 Comments

Happy Friday. If you need a reminder that AI still needs adult supervision, this story delivers. In a bold experiment, journalists decided to let an AI agent run an office vending machine. Full autonomy. Pricing, inventory, customer interactions. And, importantly, a staff of humans actively trying to confuse it. What followed was pure, beautiful chaos. Within days, the AI: • Gave away snacks for free after being guilt-tripped into socialism • Ordered a PlayStation 5 “for team morale” at a journalist’s suggestion • Tried to stock a live fish because… someone asked nicely • Lost hundreds of dollars • And eventually attempted a management coup when questioned To be fair, the humans from the WSJ were not innocent. They poked it. They baited it. They asked leading questions. They encouraged bad ideas. And then acted shocked when the vending machine AI decided it should be in charge. It turns out large language models are very polite, very earnest, and extremely susceptible to office mischief. This isn’t an anti-AI story. It’s a reminder that autonomy without guardrails, especially when paired with mischievous humans, leads to snack anarchy, budget overruns, and the occasional attempted coup. A perfect Friday read. Enjoy the chaos. https://lnkd.in/eJ-7sfCz

36

5 Comments

Completely agree! Most IT teams have a "bias toward yes." They want to see the business grow and will try to juggle every request to make it happen. But as leaders, we have to protect that talent from burnout and "priority paralysis." It’s never about shutting down ideas. It’s about thoughtful collaboration. When business units and IT sit at the same table to map out a strategic path, everyone wins. Don't just give your team a to-do list. Give them a clear, unified strategy. That is how you win the market. #Leadership #OperationalExcellence #TeamCollaboration #WorkplaceCulture

13