AI-Driven Decision Support Systems for Doctors

Access to a real-time AI decision support tool during primary care visits in Nairobi cut diagnostic errors by 16% and treatment errors by 13%, with no added harm reported. 1️⃣ This study tested “AI Consult,” an LLM-powered tool integrated into EMRs at 15 Penda Health clinics in Kenya. 2️⃣ The tool ran in the background, issuing alerts only when needed (green/yellow/red), preserving clinician autonomy. 3️⃣ Across 39,849 visits, clinicians with AI support made 16% fewer diagnostic errors and 13% fewer treatment errors, as judged by blinded physician review. 4️⃣ Estimated annually, AI Consult could prevent 22,000 diagnostic and 29,000 treatment errors at Penda alone. 5️⃣ The largest error reductions were in history-taking (32% relative risk reduction) and treatment safety (NNT = 13.9). 6️⃣ Clinicians with the tool gradually made fewer mistakes even before receiving alerts, suggesting it helped build better habits. 7️⃣ All clinicians surveyed said AI Consult improved care; 75% said the improvement was “substantial.” 8️⃣ No safety events were attributed to AI Consult, and alert fatigue was mitigated through careful interface and threshold design. 9️⃣ Uptake increased after targeted deployment strategies: coaching, peer champions, and performance feedback. 🔟 The study underscores that success came not just from the model itself, but from aligning tech design with clinical workflow. ✍🏻 Robert Korom, Sarah Kiptinness, Najib Adan, Kassim Said, Catherine Ithuli, Oliver Rotich, Boniface Kimani, Irene King’ori, Stellah Kamau, Elizabeth Atemba, Muna Aden, Preston Bowman, Michael Sharman, Rebecca Soskin Hicks, MD, Rebecca Distler, Johannes H., Rahul K. Arora, Karan Singhal. AI-based Clinical Decision Support for Primary Care: A Real-World Study. 2025. DOI: 10.48550/arXiv.2407.12986

Clinical decision-making in oncology is complex, requiring the integration of multimodal data and multidomain expertise. We developed and evaluated an autonomous clinical artificial intelligence (AI) agent leveraging GPT-4 with multimodal precision oncology tools to support personalized clinical decision-making. The system incorporates vision transformers for detecting microsatellite instability and KRAS and BRAF mutations from histopathology slides, MedSAM for radiological image segmentation and web-based search tools such as OncoKB, PubMed and Google. Evaluated on 20 realistic multimodal patient cases, the AI agent autonomously used appropriate tools with 87.5% accuracy, reached correct clinical conclusions in 91.0% of cases and accurately cited relevant oncology guidelines 75.5% of the time. Compared to GPT-4 alone, the integrated AI agent drastically improved decision-making accuracy from 30.3% to 87.2%. These findings demonstrate that integrating language models with precision oncology and search tools substantially enhances clinical accuracy, establishing a robust foundation for deploying AI-driven personalized oncology support systems. Paper and research by Dyke Ferber,  Jakob Nikolas Kather and larger team. The excerpt above is from the author's abstract

𝐀𝐈 𝐀𝐠𝐞𝐧𝐭 𝐀𝐜𝐡𝐢𝐞𝐯𝐞𝐬 𝟗𝟏% 𝐀𝐜𝐜𝐮𝐫𝐚𝐜𝐲 𝐢𝐧 𝐂𝐚𝐧𝐜𝐞𝐫 𝐃𝐢𝐚𝐠𝐧𝐨𝐬𝐢𝐬 Oncology decision-making is notoriously complex. Clinicians must integrate histopathology images, radiology scans, genetic profiles, and ever-evolving treatment guidelines to make personalized care decisions. It's a cognitive challenge that even experienced specialists find demanding. A new study by Ferber et al. in Nature Cancer shows how an autonomous AI agent tackled this complexity head-on—and the results are striking. 𝗪𝗵𝘆 𝘁𝗵𝗶𝘀 𝗺𝗮𝘁𝘁𝗲𝗿𝘀: Current AI approaches in healthcare often work in isolation—analyzing single data types or providing generic responses. But real clinical decisions require synthesizing multiple sources of evidence simultaneously, something that has remained challenging for AI systems. 𝗞𝗲𝘆 𝗶𝗻𝗻𝗼𝘃𝗮𝘁𝗶𝗼𝗻𝘀: ◦ 𝐌𝐮𝐥𝐭𝐢𝐦𝐨𝐝𝐚𝐥 𝐭𝐨𝐨𝐥 𝐢𝐧𝐭𝐞𝐠𝐫𝐚𝐭𝐢𝐨𝐧: Vision transformers detect genetic mutations directly from tissue slides, MedSAM segments tumors in radiology images, and the system queries precision oncology databases autonomously ◦ 𝐒𝐞𝐪𝐮𝐞𝐧𝐭𝐢𝐚𝐥 𝐫𝐞𝐚𝐬𝐨𝐧𝐢𝐧𝐠: The agent chains tools together—first measuring tumor growth from imaging, then checking mutation databases, then searching recent literature ◦ 𝐄𝐯𝐢𝐝𝐞𝐧𝐜𝐞-𝐛𝐚𝐬𝐞𝐝 𝐜𝐢𝐭𝐚𝐭𝐢𝐨𝐧𝐬: 75.5% accuracy in citing relevant medical guidelines, addressing the critical problem of AI hallucinations in healthcare 𝗧𝗵𝗲 𝗿𝗲𝘀𝘂𝗹𝘁𝘀: When tested on 20 realistic patient cases, the integrated system achieved 91% accuracy in clinical conclusions. Perhaps more telling: GPT-4 alone managed only 30% accuracy on the same cases—nearly a 3x improvement through tool integration. The agent successfully used appropriate diagnostic tools 87.5% of the time and provided helpful responses to 94% of clinical questions. 𝗧𝗵𝗲 𝗯𝗶𝗴𝗴𝗲𝗿 𝗽𝗶𝗰𝘁𝘂𝗿𝗲: This isn't about replacing oncologists—it's about augmenting clinical reasoning with AI that can process multiple data streams simultaneously. The modular approach means individual tools can be updated, validated, and regulated independently. While challenges remain around data privacy and regulatory approval, this research points toward a future where AI agents serve as sophisticated clinical reasoning partners, helping doctors navigate the increasing complexity of modern medicine. https://lnkd.in/e52xBZj9 #AIinHealthcare #PrecisionOncology #ClinicalAI #DigitalHealth #MachineLearning #Oncology

The time to design AI-native architectures isn’t after operational gaps appear. It’s now. Healthcare doesn’t need more AI pilots. It needs systems that can reason, coordinate, and decide — together, in real time. On that line of thought, sharing this recent peer-reviewed commentary by Dr. Andrew Borkowski that outlines how multiagent AI systems are reshaping the frontier of clinical intelligence. These systems go far beyond today’s static tools and LLM wrappers. They orchestrate collaboration — across agents, workflows, and decision points. The commentary shares an example of sepsis management, where seven AI agents work in parallel to: • Clean and integrate unstructured data • Interpret imaging and vitals via deep learning • Stratify risk with Sequential Organ Failure Assessment (SOFA) and qSOFA scores • Generate treatment plans using reinforcement learning • Optimize hospital logistics with queue theory and genetic algorithms • Detect anomalies in real time via streaming forecasts • Auto-document every step into structured EHR records Every decision is governed by explainable AI, a quality-control agent, and confidence-calibrated outputs. Federated learning enables continuous evolution, while blockchain and OAuth 2.0 protect system integrity. This isn’t a distant vision. It’s a working blueprint for health systems under pressure to scale intelligence, not just automation. 📌 Read the commentary here → https://lnkd.in/g5X5PADk #AIsystems

In a blinded clinical trial, an AI doctor planned better treatments than real physicians! Meet AMIE (Articulate Medical Intelligence Explorer) - Google DeepMind's new multi-agent clinical reasoning system. In a randomized, blinded OSCE - the gold-standard test used to license physicians, AMIE went head-to-head with 21 board-certified doctors across 100 patient cases spanning cardiology, neurology, pulmonology, gastroenterology, and OB/GYN. Each case had three visits, meaning AMIE had to remember prior symptoms, track progress, and adapt therapy just like a real doctor. The results: → Overall management quality: 88% AMIE vs 74% physicians (p = 0.019) → Treatment precision: 94% vs 67% on visit one → Investigation precision: 99–100% vs 84–88% → Guideline alignment: 89–93% vs 75–81% → Human preference: specialists and patients chose AMIE 42% vs 8% → Zero domains where doctors outperformed it And this wasn’t a Q&A test. It was a live, longitudinal exam of clinical reasoning, communication, and empathy. So how did AMIE do it? By splitting the work like a real care team. → The Dialogue Agent talks with patients - fast, empathetic, and memory-rich. → The Management Reasoning Agent reads NICE and BMJ guidelines in real time and builds structured, evidence-based care plans. One agent thinks fast. The other thinks deep. Together, they reason safely and explain every step. That’s the shift: from single, opaque chatbots → to multi-agent clinical systems that plan, remember, and justify. Medicine is no longer a one-answer problem. It’s a reasoning problem and AMIE just proved AI can handle it.

Earlier this year, a close family member was dangerously ill in India. The diagnosis wasn’t working. The symptoms were escalating. No one knew why. We felt that familiar dread - being far from the situation, and even farther from certainty. So I did what millions of people now do in moments of uncertainty: I asked ChatGPT. Typed in the symptoms, context, and history - not expecting magic, just hoping for perspective. What came back was startlingly precise: It could be this. If so, check the kidneys. If the kidneys are involved, watch for infection. If it’s in the blood, it could be sepsis. Escalate - fast. It was right. All of it. We flagged it to the doctors. It shaped the next set of tests. And it helped turn a very bad situation around- fast. That moment crystallized something for me: AI isn’t about replacing doctors. It’s about replacing helplessness. There’s a lot of talk in Silicon Valley about curing death. Like, literally - curing aging, reversing entropy, building new bodies from cells that forgot they were old. Some of it will work. Much of it will take decades. But the more immediate, life-changing breakthroughs are already happening - not at the edge of life, but at the frontlines of medicine. Just this week: ▪️Microsoft AI Diagnostic Orchestrator (MAI‑DxO) outperformed experienced physicians. It was tested on 304 real-world case studies published in The New England Journal of Medicine. MAI-DxO solved 85.5% of them. By comparison, 21 experienced physicians solved just 20%. How? By mimicking a panel of clinical minds: One AI model orders tests, another evaluates the results, others debate, reframe, escalate. It’s a structured, chain-of-thought system modeled on real diagnostic reasoning. And it recommended fewer unnecessary tests, saving both time and cost. Yes, it’s early. It hasn’t been deployed in hospitals. But the signal is loud: we’re not far from AI-powered co-pilots for frontline care. ▪️ Google DeepMind's AlphaGenome, tackled a different frontier: the "dark matter" of DNA. Most disease-causing mutations don’t lie in genes, they hide in the regulatory code. Until now, we couldn’t see them at scale. AlphaGenome can process 1m base pairs at once - entire genomic neighborhoods. It’s already predicted how some non-coding mutations can trigger cancer. And it trained in just four hours. If MAI‑DxO gives us a better map of what’s happening now, AlphaGenome gives us a telescope into what might happen next. These tools don’t just answer questions. They reshape who gets to ask them. This is what makes the AI revolution in medicine so powerful. Not just that it might one day extend life. But that it already extends understanding. That it makes complexity legible. That it turns patients into partners - and doctors into augmented super-thinkers. And that alone could save millions of lives.

Did you see the recent news??? Microsoft recently unveiled its latest AI Diagnostic Orchestrator (MAI DxO), reporting an impressive 85.5% accuracy on 304 particularly complex cases from the New England Journal of Medicine, compared to just ~20% for physicians under controlled conditions . These results—quadrupling the diagnostic accuracy of human clinicians and more cost-effective than standard pathways — have gotten a lot of buzz. They may mark a significant milestone in clinical decision support and raise both enthusiasm but also caution. Some perspective as we continue to determine the role of AI in healthcare. 1. Validation Is Essential Promising results in controlled settings are just the beginning. We urge Microsoft and others to pursue transparent, peer reviewed clinical studies, including real-world trials comparing AI-assisted workflows against standard clinician performance—ideally published in clinical journals. 2. Recognize the value of Patient–Physician Relations Even the most advanced AI cannot replicate the human touch—listening, interpreting, and guiding patients through uncertainty. Physicians must retain control, using AI as a tool, not a crutch. 3. Acknowledge Potential Bias AI is only as strong as its training data. We must ensure representation across demographics and guard against replicating systemic biases. Transparency in model design and evaluation standards is non-negotiable. 4. Regulatory & Liability Frameworks As AI enters clinical care, we need clear pathways from FDA approval to liability guidelines. The AMA is actively engaging with regulators, insurers, and health systems to craft policies that ensure safety, data integrity, and professional accountability. 5. Prioritize Clinician Wellness Tools that reduce diagnostic uncertainty and documentation burden can strengthen clinician well-being. But meaningful adoption requires integration with workflow, training, and ongoing support. We need to look at this from a holistic perspective. We need to promote an environment where physicians, patients, and AI systems collaborate, Let’s convene cross sector partnerships across industry, academia, and government to champion AI that empowers clinicians, enhances patient care, and protects public health. Let’s embrace innovation—not as a replacement for human care, but as its greatest ally. #healthcare #ai #innovation #physicians https://lnkd.in/ew-j7yNS

We’re excited to share research from our health team at Microsoft AI: a proof-of-concept showing that AI can master medicine’s most intricate diagnostic challenges by following the same step-by-step reasoning expert physicians use. There's more detail in our pre print paper & blog Paper-> https://lnkd.in/egDiNsqR Blog-> https://lnkd.in/esGFhSeB Sharing what I'm most excited about from this work. 1. Benchmarks Traditional medical benchmarks like the USMLEs condense clinical cases into neat multiple-choice questions—far from the real clinical workflow. We’ve approached things in a different way: Sequential Diagnosis Benchmark (SDBench) deconstructs 304 of the most diagnostically complex and demanding cases in medicine published in the New England Journal of Medicine. SD Bench requires models—and physician—to begin with an initial presentation, ask follow-up questions, order tests, and converge on the confirmed diagnosis—just as in routine clinical practice. You can see how this works in a video with Xiao Liu on our blog. 2. Performance With this new benchmark we tested a suite of the best known generative AI models against the 304 NEJM cases with impressive out-the-box performance. Beyond this we developed the Microsoft AI Diagnostic Orchestrator (MAI-DxO). By emulating a virtual panel of physicians with diverse thinking styles, MAI-DxO boosts raw model accuracy and solves a remarkable 85.5% of NEJM cases. For comparison we evaluated 21 practicing UK/US physicians and on the same tasks, these experts achieved a mean accuracy of 20%. 3. Costs One of our concerns was that AI would default to ordering every investigation to arrive at the correct diagnosis. So we set the system up such that each requested investigation also incurred a cost. This allowed us to evaluate performance against both diagnostic accuracy and resource expenditure. As MAI-DxO is configurable it is seen to operate along a Pareto frontier of accuracy versus resource use. What’s next: While for now exciting research, we believe this kind of superhuman clinical reasoning will in future reshape medicine. A particular focus for our group is on consumer health. Today, Bing and Copilot answer over 50 million health queries daily—from a first-time knee-pain search to finding a late-night pharmacy. We’re committed to bringing rigorous and reliable AI support into these journeys, backed by clinical evidence and robust commitments to quality, safety and trust. A huge shout-out to everyone on our new team who contributed, our partners across Microsoft, and particularly to Mustafa Suleyman for his vision. He saw the opportunity for AI to improve healthcare more than a decade ago and it now feels like this is the right time to deliver on the opportunity. Harsha Nori Mayank Daswani Christopher Kelly Scott Lundberg Marco Túlio Ribeiro Marc Wilson Xiao Liu Viknesh Sounderajah Bay Gross Peter Hames Eric Horvitz Charlotte Cooper Simpson, PhD

Researchers at Harvard Medical School have developed an AI system called Dr. CaBot that provides detailed diagnostic reasoning for challenging medical cases. Unlike most AI diagnostic tools, Dr. CaBot explains its thought process step-by-step, generating a comprehensive list of possible diagnoses (differential diagnosis) and narrowing down to a final answer, similar to how expert physicians approach complex cases. For the first time, the prestigious New England Journal of Medicine (NEJM) published an AI-generated diagnosis by Dr. CaBot alongside a human expert’s opinion for a real medical case. This highlights Dr. CaBot’s usefulness as a medical education and research tool. Dr. CaBot can: Present narrated, lifelike video explanations with detailed reasoning, Search millions of clinical abstracts to provide citations and avoid factual errors, Analyze thousands of historical medical cases to emulate expert reasoning. Dr. CaBot is not yet ready for clinical use but is being demonstrated at Boston-area hospitals for feedback. The developers note Dr. CaBot’s speed and availability—compared to human clinicians—and see promise for educational, research, and potentially practical uses with further refinement.

𝗔𝗜 𝗔𝗴𝗲𝗻𝘁𝘀 𝗖𝗼𝘂𝗹𝗱 𝗧𝗿𝗮𝗻𝘀𝗳𝗼𝗿𝗺 𝗖𝗹𝗶𝗻𝗶𝗰𝗮𝗹 𝗗𝗲𝗰𝗶𝘀𝗶𝗼𝗻 𝗦𝘂𝗽𝗽𝗼𝗿𝘁. 𝗛𝗲𝗿𝗲'𝘀 𝗪𝗵𝗮𝘁 𝗧𝗵𝗮𝘁 𝗠𝗶𝗴𝗵𝘁 𝗟𝗼𝗼𝗸 𝗟𝗶𝗸𝗲. This week, OpenAI released a visual tool for building multi-agent workflows(1). I've been curious about agents for a while, but never had time to learn. Playing with this tool got me thinking about a longstanding challenge: how do we translate complex clinical pathways into effective CDS tools? 𝗧𝗵𝗲 𝗣𝗿𝗼𝗯𝗹𝗲𝗺 𝘄𝗶𝘁𝗵 𝗖𝘂𝗿𝗿𝗲𝗻𝘁 𝗖𝗗𝗦 Today's EHR-based decision support faces a fundamental tradeoff. Tightly scripted rules are reliable but inflexible. Loosely scripted ones give clinicians room to adapt but sacrifice consistency. Multi-agent AI workflows might offer a way out of this bind. 𝗔 𝗣𝗿𝗼𝗼𝗳 𝗼𝗳 𝗖𝗼𝗻𝗰𝗲𝗽𝘁 To explore this, I translated Children's Hospital of Philadelphia's Suicide Risk Assessment Pathway (2) into a multi-agent workflow. Here's how it works: • Input: Clinician's risk formulation, screening results, risk and protective factors • Acuity script: Determines patient acuity level • Intervention agent: Recommends response level • Response agents: Four specialized agents provide tailored clinical guidance based on severity 𝗔 𝗗𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝗞𝗶𝗻𝗱 𝗼𝗳 𝗖𝗗𝗦 Now imagine this in your EHR. Instead of rigid decision trees, you'd have multiple specialized LLMs that can: • Access relevant patient data • Provide contextual guidance • Answer follow-up questions in real time • Adapt to clinical nuance while maintaining evidence-based standards Some extras that make this promising are the ability to use MCP and RAG! This isn't just automation. It's augmentation that preserves clinical judgment while providing robust support.   Bimal Desai MD, MBI, FAAP, FAMIA