Clinical Decision Support

Oncology AI reading the full patient chart, founded 2021 on the observation that while software could already handle the roughly 20 percent of medical data that is structured, generative AI made the other 80 percent, the free text notes, tractable. OncoLLM is the underlying platform and the architecture is disclosed with unusual specificity: not one model but a constellation of eight specialized models ranging from 3 billion to 72 billion parameters, working agentically to interpret information at the patient level. Prism is the trial matching application, screening patients with upcoming appointments against all active trials and working in both directions, patient to trial and trial to patient, with continuous tracking. Matches are cited back to pathology, biomarkers, and note level evidence rather than returned as a bare recommendation. The platform has expanded beyond matching into verifiable patient summaries for visit preparation, cancer registry abstraction, and structured data generation for research. Reported results include a 40 percent increase in trial matches, more than 30 percent increase in enrollments, and 67 percent reduction in chart review time, with one NCI designated center reporting 100 percent screening coverage at three times the throughput of manual review. Peer reviewed validation has appeared in Nature Digital Medicine with presentation at ASCO, and a pilot study at the Medical College of Wisconsin Cancer Center covers gastrointestinal, genitourinary, breast, and thoracic teams. Mount Sinai Tisch Cancer Center deployed Prism in January 2026, becoming the first NCI designated Comprehensive Cancer Center in New York City to use it for systemwide matching. More than $36 million raised; co-founders Sarim Khan and Hrituraj Singh.

Note on naming: this is Artera of Los Altos, California (artera.ai), developer of multimodal AI cancer tests. It is unrelated to the separately indexed Artera of Santa Barbara (artera.io), which makes patient communication agents. The two companies share a name and nothing else. The ArteraAI Prostate Test is a multimodal artificial intelligence model combining digitized histopathology images from an existing biopsy with structured clinical data including age, PSA, and tumor stage, producing a continuous risk score from 0 to 1 with pre established cut points corresponding to roughly 3 percent and 10 percent estimated ten year risk of distant metastasis. It requires no additional procedure because it reads tissue already taken. Beyond prognosis it is predictive: for NCCN intermediate risk patients it estimates whether adding short term androgen deprivation therapy to radiation will reduce risk, which the literature describes as the first predictive biomarker for that decision. The algorithm is pre established and locked at version 1.2, developed from large datasets and multiple phase 3 randomized trials with up to 15 years of follow up. Regulatory and guideline standing is unusually complete: FDA De Novo marketing authorization in August 2025, inclusion in the NCCN Clinical Practice Guidelines for Prostate Cancer as the first and only AI risk stratification tool, and CMS coverage with a payment rate effective January 2024. In June 2026 the company introduced a digital pathology based test providing individual risk estimates in metastatic hormone sensitive prostate cancer. Co-founded by Felix Y. Feng of UCSF; CEO Andre Esteva.

AI clinical platform built for Latin America, and the first vendor in this index whose primary market is outside the United States and Europe. The platform combines ambient documentation, clinical decision support, and EHR integration: it transcribes consultations in real time and generates structured records adapted to each clinician's style, reviews the record, flags potential errors, and surfaces evidence based recommendations and institutional protocol guidance during the encounter. Models are trained on clinical guidelines, peer reviewed literature, and local institutional protocols, which is the substantive localization claim rather than translation alone. Since launching July 2025 the company reports deployment across more than 25 hospital systems in Brazil, Colombia, and Mexico, reaching 14 million patients and processing over 5 million consultations. Reported outcomes at customer sites include protocol adherence rising from 84 percent to 99 percent, physicians recovering roughly 1.7 to 2 hours daily, and 60,000 medical errors flagged in real time. Mater Dei in Brazil reports physicians using the platform around eight hours a day. A free tier targets independent private practice physicians who lack institutional technology access. The company positions the product as supporting clinicians rather than deciding, with the physician making the final call, and is operating while AI regulation is still forming regionally: Brazil's Senate has approved a risk based AI bill pending further approval, and Colombia has sent its own bill to Congress. Founded by Nicolas Abad after his father, a physician, died from a preventable drug interaction. $42 million raised including a $33 million Series A led by Andreessen Horowitz.

Indexed for what is commercially available today, which requires care because the company's own AI algorithms are not yet cleared. MyoVista Insights is a cloud based ECG reporting and management platform, in phased rollout since May 2025 and shipping at version 1.3 as of June 2026, positioned to replace on premise ECG management software. Its AI-ECG Algorithm Marketplace, introduced in version 1.3, integrates third party algorithm results into the platform, and the first algorithm made available is Bunkerhill Health's FDA cleared ECG-EF model detecting reduced left ventricular ejection fraction at or below 40 percent from a routine 12-lead ECG. Bunkerhill is separately indexed here, so this is a distribution relationship between two vendors in this index rather than a HeartSciences algorithm. The company's own pipeline is not yet available: the MyoVista wavECG device, a 12-lead resting ECG built to host AI-ECG algorithms, was submitted for 510(k) clearance in December 2025 and clearance was not confirmed at the time of this record; its impaired cardiac relaxation algorithm was deliberately separated from that submission to complete additional validation against revised American Society of Echocardiography guidelines for diastolic dysfunction; and its low ejection fraction algorithm remains in pre submission work. Underlying intellectual property is licensed in part from the Icahn School of Medicine at Mount Sinai. Assistive AI-ECG assessments are eligible for Medicare reimbursement under the Hospital Outpatient Prospective Payment System at APC 5734. NASDAQ: HSCS.

AI utilization management, and the only vendor in this index positioned to sit between payer and provider rather than serving one side. The Dragonfly platform (formerly CORTEX) uses real time predictive analytics to continuously assign an objective medical necessity score, the proprietary Care Level Score, and an anticipated level of care for every patient, with the stated intent of removing subjectivity from patient status determination so both parties work from the same evidence. Product lines cover Utilize for utilization review and case management, Navigate for discharge readiness and length of stay, Advise for physician advisor workflow and peer to peer escalation, and revenue integrity for denials and appeals. Precision UM is the deepest tier, a shared utilization management approach between a health system and a health plan built on common clinical data and historical determination patterns. Integration with the EMR is bidirectional. The company also sells Physician Advisor Services, supplementing night and weekend coverage or operating fully outsourced, so buyers should separate the platform from the staffed service. Named outcome: AnMed reported 14.6x return on investment over a twelve month period and 1,221 hours of front end review time saved through Precision UM with a national health plan. A third party study found Dragonfly rendered clinical review determinations 38 percent faster than fax and 15 percent faster than through the EMR. Headquartered in Franklin, Tennessee; co-founded by Joan Butters.

Indexed for the AI platform and agent portfolio rather than for the wider clinical research services, data, and consulting business, which is context under this index's product scoping rule. IQVIA (NYSE: IQV) launched IQVIA.ai in March 2026, a unified agentic platform and agent marketplace spanning clinical, commercial, and real world operations, built on the company's proprietary health data and Healthcare grade AI together with NVIDIA Nemotron, NeMo Agent Toolkit, Dynamo, and LangChain, following a collaboration begun over a year earlier. Disclosed agent use cases include target identification, clinical data review, literature review, market assessment, and healthcare professional engagement. The company reports filing more than 100 AI related patents and deploying more than 150 intelligent agents across internal teams and client environments, with 19 of the top 20 pharmaceutical companies having begun incorporating IQVIA agents into their workflows. IQVIA AI Assistant is a generative interface over the company's existing solutions and data products. The commercial side of the portfolio includes healthcare professional targeting and engagement, which sits closer to pharmaceutical marketing than to clinical operations and should be evaluated separately from the research and clinical agents.

Real world evidence generation, spun out of Stanford's Green Button project in 2020 and built to answer the question a clinician actually has at the bedside when no trial covers their specific patient. GENEVA OS is the underlying operating system across a federated network the company describes as exceeding 300 million anonymized patient records. ChatRWD is a chat based application that generates publication grade observational studies in minutes rather than months, without requiring a data scientist in the loop, and Green Button remains as a consultation service delivering evidence within 48 hours. Alexandria is the company's evidence library of pre computed evidence briefs. Two design decisions distinguish it: ChatRWD is an LLM independent framework, so an institution can maintain its own model security posture, and every answer carries a Real World Fitness Score rating how fit for purpose the underlying dataset was for that specific question, which is an explicit and unusual admission that evidence quality varies by question. Stanford Health Care has moved from using the service to embedding it in physician workflow and clinical notes. The Atropos Evidence Agent is distributed via the Databricks Marketplace under a partnership announced June 2025.

Clinical intelligence platform sold to health plans, spanning utilization management, payment integrity, appeals, care management, and policy management. The company states it is sold to health plans, and that providers and manufacturers do not license the platform, which makes it payer side alongside Alaffia. Cohere Unify combines a reported 350 or more clinically trained AI models with workflow automation and human review to auto determine prior authorization requests in real time, and Cohere Connect provides the prior authorization APIs, reported to have carried more than 15 million submissions and to support 47 million payer provider interactions annually. The critical design fact is the direction of automation: the company reports up to 85 percent real time approvals and states explicitly that remaining submissions are reviewed by a clinician before final determination, meaning the model approves and humans decide the rest. Additional products include Cohere Review Assist for acute inpatient care, Cohere Policy Studio, and a Payment Integrity Suite extended through the September 2025 acquisition of ZignaAI. Reported outcomes include care access 70 percent faster than traditional processes, up to 9x return on investment, 94 percent provider satisfaction, and a provider NPS of 67. Named plan relationships include Humana and Geisinger. $90 million Series C in May 2025.

Clinical AI platform built around Almanac Copilot, which the company describes as a Level 1 autonomous EHR agent: it retrieves and summarizes patient data, drafts notes, places orders, and surfaces prioritized alerts, dynamically selecting and chaining tools from a defined set including FHIR functions, a browser, clinical calculators, a Python interpreter, and the EHR database. As a Level 1 agent it acts only on explicit clinician command and requires review and approval of every action. Grounded in peer reviewed evidence in a vector database rather than an unconstrained model, and stated to be free of pharmaceutical advertising, a deliberate contrast with ad supported clinical reference products. On the EHR QA benchmark of 300 common EHR tasks it reported a 74 percent completion rate, matching much larger models. Founded by physician researcher Cyril Zakka MD, whose Stanford work introducing retrieval augmented generation to clinical medicine became one of NEJM AI's most cited papers. Important scope note carried from the source record: the company characterizes the platform as validated through research rather than shipped as a finished product, and it is undergoing clinical validation in academic medical center settings. $10 million seed in April 2026 led by F-Prime with General Catalyst and Lightspeed.

Clinical AI platform pairing ambient scribing with evidence grounded clinical reasoning, which the other scribes do not attempt. Clinicians describe a case in natural language and the platform returns a ranked differential with suggested workups, drafts a problem oriented assessment and plan with inline evidence links, and answers clinical questions with citations drawn from an index of more than 38 million peer reviewed articles, consensus guidelines, and FDA information covering over 154,000 drug compounds. In the Harvard and Stanford ARISE NOHARM evaluation, a physician validated clinical safety benchmark, the platform ranked among the top medical AI systems. The scribing side produces clinic notes, H&P notes, progress notes, discharge summaries, and patient handouts, with assisted EHR workflows across Epic, eClinicalWorks, athenahealth, and Elation. A Developer API exposes the full stack, including clinical question answering, chart summarization, differential generation, care planning, and scribing, so other platforms can embed the clinical intelligence. Independent testing notes reasoning is strongest on common presentations and weaker on complex multi system cases, and the company positions the product as decision support requiring physician judgment on every output. Founded 2021 by Dr Dereck Paul and Graham Ramsey.

Clinical AI for imaging triage, delivered through aiOS, an enterprise AI operating system that handles data normalization, continuous performance monitoring, and governance so health systems can run multi condition AI without re architecting infrastructure. In January 2026 the FDA cleared what the company describes as healthcare's first comprehensive AI triage solution, powered by CARE (Clinical AI Reasoning Engine), its self developed foundation model. The clearance brought 11 newly cleared indications together with three previously cleared into a single abdomen CT workflow, 14 in total, and per the FDA reviewed pivotal study the new indications reported a mean sensitivity of 97 percent and mean specificity of 98 percent, with the company reporting roughly an order of magnitude reduction in false alerts against single condition tools. The company reports more than 100 million patient cases analyzed on aiOS. A successor model, CARE2, has been announced, and a multi year collaboration with Amazon Web Services supports the foundation model work. Co founded and led by Elad Walach.

AI platform that retrieves, aggregates, and structures fragmented medical records into longitudinal patient histories, originally built for oncology and since extended to all therapeutic areas. The Clinical Clarity Engine applies natural language processing and machine learning to unstructured records; the company reports processing more than 300 million medical records from over 550,000 healthcare locations, with every structured output linked back to its source document. The xDECIDE provider portal presents structured data, autogenerated summaries, and AI assisted treatment option reports with scientific rationale, ranked by the xCORE engine and reviewed through a virtual tumor board; xINFORM is the patient facing portal. Data products supply longitudinal real world oncology datasets to research and regulatory programs. Holds HITRUST e1 certification. Raised a $46 million Series B in June 2026.

Agentic AI platform for health systems. Carebricks lets clinical and operational teams build, deploy, and govern their own AI agents across clinical, operational, and administrative workflows rather than buying a point solution per use case. Agents in production include coronary calcium detection on routine chest CT, nephrology triage, lung nodule follow up, referral prioritization, prior authorization packet assembly, and registry automation. The company develops its own FDA cleared imaging algorithms that run inside the platform, including algorithms for coronary artery calcium and aortic valve calcium on contrast enhanced non gated chest CT, mitral annular calcification, and bone mineral density. CMS established a national billing code and OPPS payment for algorithmic CAC and AVC analysis on chest CT effective April 1, 2026. In production at Cleveland Clinic, the University of Texas Medical Branch, Intermountain Health, and Mayo Clinic. Founded by Nishith Khandwala, previously a researcher at Stanford's Center for Artificial Intelligence in Medicine and Imaging.

Precision medicine company (NASDAQ: TEM) whose provider software is what this index covers: Next, an AI clinical decision support system that alerts clinicians to patients who may have fallen off care guidelines, backed by a vendor reported 662 patient prospective study across six sites; Hub, a physician platform rebuilt on an agentic architecture including a prior authorization agent; and David, a generative AI clinical assistant deployed directly into the EHR, with Northwestern Medicine as the first health system. Tempus acquired Deep 6 AI in March 2025, adding a precision research platform that applies natural language processing to structured and unstructured EMR data to match patients to clinical trials in near real time and to generate real world evidence. Deep 6 reports real time EMR feeds across more than 30 health systems, an ecosystem of over 1,000 research facilities including 18 academic medical centers and 11 NCI designated cancer centers, and that sites find more than 25 percent more patients than with traditional recruitment; the company notes 92 percent of trial inclusion and exclusion criteria benefit from unstructured data and that 15 to 20 percent of eligible patients are found through unstructured data alone. Outputs are positioned as decision support requiring trial team validation before action. The sequencing and pharma services businesses are context, not indexed products.

AI medical search and clinical decision support used at the point of care, free to verified United States clinicians and monetized through pharmaceutical advertising, with enterprise health system deployments embedding the platform in the EHR. The company reports more than 757,000 verified physician users and use by more than 40 percent of practicing US physicians, with enterprise per seat contracts emerging alongside the free tier.