IBM Watson Health
| File:Merative logo.svg | |
| Company type | Private |
|---|---|
| Predecessor | IBM Watson Health |
| Founded | June 30, 2022 |
| Headquarters | Ann Arbor, Michigan |
Key people |
|
| Owner | Francisco Partners |
Number of employees | 1,000–5,000 |
| Website | merative |
Merative L.P., previously known as IBM Watson Health, is a U.S.-based health technology firm delivering solutions that facilitate medical research, clinical research, real world evidence, and healthcare operations through the use of artificial intelligence, data analytics, cloud computing, and sophisticated IT systems. Owned by the San Francisco private equity group Francisco Partners, Merative was separated from IBM's Watson Health division in 2022. As of 2023, it functions independently with its headquarters in Ann Arbor and innovation hubs located in Ireland, Hyderabad, Bengaluru, and Chennai.
History
Thomson Healthcare, originally a segment of the Thomson Corporation, became part of Thomson Reuters after their merger in 2008. On April 23, 2012, Thomson Reuters divested this unit to Veritas Capital for $1.25 billion. The deal finalized on June 6, 2012, resulting in the establishment of Truven Health Analytics, which concentrated solely on healthcare.
On February 18, 2016, IBM acquired Truven Health Analytics for $2.6 billion and integrated it into its Watson Health division. Truven specialized in comprehensive healthcare data and analytics services; its name is derived from the words "trusted" and "proven."
In January 2022, IBM sold a portion of Watson Health's assets—including Truven—to Francisco Partners for approximately $1 billion. The acquisition was completed on June 30, 2022, when Francisco Partners launched Merative as a dedicated healthcare data enterprise.
Advancements
Watson’s capabilities in natural language processing, hypothesis generation, and evidence-based learning are being investigated for use in clinical decision support systems, advancing the role of artificial intelligence in healthcare for medical practitioners. When clinicians input patient symptoms and relevant factors, Watson extracts critical details, analyzes patient and genetic information, formulates and evaluates hypotheses, then provides tailored treatment recommendations with confidence scores. Its data inputs include treatment protocols, electronic medical records, provider notes, scientific literature, clinical trials, journal publications, and patient data. Though promoted as a "diagnosis and treatment advisor," Watson’s function is limited to supporting treatment option identification post-diagnosis.
In February 2011, IBM collaborated with Nuance Communications to create a commercial offering utilizing Watson’s clinical decision support features. Physicians from Columbia University and the University of Maryland partnered to identify key medical challenges and refine Watson’s interaction with healthcare providers.
By September 2011, IBM had formed a partnership with WellPoint (now known as Anthem) to apply Watson’s analytical capabilities toward suggesting treatment plans. The first commercial application occurred by February 2013 at the Memorial Sloan–Kettering Cancer Center, where Watson was used in utilization management for lung cancer therapies.
IBM also teamed with the Cleveland Clinic in October 2012 to deploy Watson at the Cleveland Clinic Lerner College of Medicine of Case Western Reserve University, aimed at enhancing clinical expertise and improving patient care through efficient processing of extensive datasets.
In 2013, IBM and the MD Anderson Cancer Center initiated a pilot program focused on accelerating cancer treatment advancements; however, after investing $62 million without achieving objectives, the initiative was terminated.
On February 8, 2013, oncologists at the Maine Center for Cancer Medicine along with Westmed Medical Group in New York began evaluating Watson’s recommendations for lung cancer treatments.
On July 29, 2016, IBM launched IBM Watson for Oncology at Manipal Hospitals in India—the first hospital worldwide to offer personalized evidence-based cancer care insights digitally through this platform. Manipal ended this partnership in December 2018.
On January 7, 2017, IBM signed an agreement with Fukoku Mutual Life Insurance to implement IBM Watson Explorer AI for analyzing compensation payouts. This automation led to a reduction of 34 jobs while boosting productivity by 30% and generating savings of ¥140 million in operational expenses.
IBM Watson is designed to encapsulate the knowledge equivalent to that of 1,000 cancer specialists and represents a disruptive innovation within healthcare; nevertheless, its oncology applications are still emerging.
Healthcare startups have adopted seven distinct business model archetypes inspired by IBM Watson solutions to capture value by targeting either patients or healthcare providers/payers through information provision or stakeholder connectivity.
By 2019, industry assessments characterized the Watson Health initiative as an example cautioning against overstating AI capabilities. A 2021 report by the Association for Computing Machinery (ACM) highlighted challenges in portfolio management following multiple acquisitions during Watson Health’s expansion phase starting in 2015 and criticized its prioritization of branding over addressing customer requirements.
On January 21, 2022, IBM announced it would sell its Watson Health assets to private equity firm Francisco Partners.
Industry factors and obstacles
The consolidation of major healthcare corporations facilitates greater access to health data, promoting wider adoption of AI technologies. This merging strategy is designed to harness extensive datasets to enhance clinical decision support tools, a central focus within healthcare AI development. As the quantity of data expands, AI systems improve in effectiveness, prompting numerous organizations to pursue big data opportunities in the medical sector.
IBM’s Watson Oncology is being developed in partnership with Memorial Sloan Kettering Cancer Center and Cleveland Clinic. Additionally, IBM collaborates with CVS Health to apply AI in managing chronic illnesses and works alongside Johnson & Johnson to extract insights from scientific publications for drug discovery. In 2017, IBM joined forces with Rensselaer Polytechnic Institute on the Health Empowerment by Analytics, Learning and Semantics (HEALS) initiative, which leverages AI to advance healthcare results.
Similarly, other leading enterprises have committed resources to developing AI algorithms for healthcare purposes, illustrating a widespread industry movement toward incorporating sophisticated analytics and cognitive computing into medical services.
Microsoft
Microsoft’s Hanover initiative, conducted in collaboration with Oregon Health & Science University’s Knight Cancer Institute, utilizes artificial intelligence to interpret medical research and forecast effective cancer treatment options. Additional projects concentrate on analyzing medical images related to tumor development and engineering programmable cells.
Google’s DeepMind platform collaborates with the UK’s National Health Service (NHS) to identify health risks by analyzing data from mobile applications. Another NHS-backed project employs computer vision techniques on medical imaging to detect cancerous tissue.
Intel
Intel Capital, the investment branch of Intel, has funded Lumiata, a startup that uses AI to predict patients at risk and formulate care plans.
Artificial intelligence in healthcare applies sophisticated algorithms that mimic human reasoning when interpreting medical information, enabling conclusions without direct human intervention. Unlike conventional technologies, AI utilizes machine learning models that detect patterns and produce logical outcomes. These models undergo rigorous validation to reduce errors and function both literally—executing explicit instructions—and as black box systems that deliver accurate predictions without revealing their internal rationale.
AI tools mainly focus on analyzing correlations between preventive or therapeutic measures and patient results. They assist in areas such as diagnosis, treatment protocol formulation, drug development, personalized medicine approaches, and continuous patient monitoring. Prestigious institutions like The Mayo Clinic, Memorial Sloan Kettering Cancer Center, and the NHS have created AI algorithms for clinical applications. Prominent technology companies such as IBM and Google, along with startups including Welltok and Ayasdi, also develop AI-driven solutions.
Healthcare facilities increasingly implement AI-powered predictive analytics to improve operational workflows, reduce expenses, enhance patient experience, and optimize workforce deployment. These technologies support healthcare administrators in raising utilization rates, minimizing patient boarding durations, shortening hospital stays, and refining staff management.
The following medical specialties attract significant attention within artificial intelligence research:
Radiology
The integration of AI in radiology enhances the identification of subtle imaging abnormalities that may elude clinicians. Research from Stanford introduced an algorithm capable of detecting pneumonia with a superior average F1 score compared to radiologists involved in the study. The Radiological Society of North America now includes AI-focused imaging sessions at its yearly conference. Despite measurable improvements in specific performance metrics, some experts regard AI as a challenge due to its variable results in isolated comparisons against seasoned radiologists.
Imaging
AI technologies have found applications in assessing outcomes following orthognathic surgery and cleft palate treatments, as well as evaluating facial attractiveness and perceived age. A 2018 investigation published in the Annals of Oncology revealed that a deep learning convolutional neural network identified skin cancer with 95% accuracy, surpassing dermatologists who achieved 86.6%. This underscores the expanding influence of AI in diagnostic imaging fields.
Disease Diagnosis
Artificial intelligence has been successfully utilized for diagnosing critical illnesses such as Diabetes and Cardiovascular Disease (CVD), both among the top global mortality contributors. Due to their significant fatality rates, extensive research efforts have focused on applying AI techniques to enhance diagnostic precision.
Jiang et al. (2017) reviewed various AI approaches employed across different diseases, including Support vector machines, neural networks, and decision trees, all designed to improve classification accuracy aligned with specific training objectives.
In particular, two AI methodologies—Artificial Neural Networks (ANN) and Bayesian Networks (BN)—are widely recognized for classifying diabetes and CVD. A literature review spanning 2008–2017 concluded that early disease identification is feasible using these machine learning frameworks. Among them, ANN demonstrated superior performance over BN, achieving average accuracies of 87.29% for diabetes and 89.38% for cardiovascular disease.
Telehealth
The expansion of Telemedicine has broadened AI’s role in patient surveillance. AI-powered platforms can notify healthcare providers about possible disease progression by analyzing live data streams. Wearable technology supports continuous monitoring, capturing subtle physiological variations often overlooked by clinicians, thereby strengthening remote healthcare delivery.
Electronic health records
Electronic health records (EHRs) serve as foundational components for healthcare digitization and information exchange. Nevertheless, the extensive documentation required can lead to cognitive strain and professional burnout. To mitigate these challenges, EHR systems are increasingly incorporating automation and natural language processing (NLP) capabilities. Research conducted by the Centerstone Research Institute demonstrated that predictive models leveraging EHR data can forecast personalized treatment outcomes with 70–72% accuracy, illustrating AI’s potential to identify disease etiologies from clinical records.
Drug Interactions
Progress in natural language processing has facilitated the development of algorithms capable of identifying drug-drug interactions within scientific literature. These interactions present considerable risks for patients on polypharmacy regimens, with danger escalating alongside the number of medications prescribed. To address this complexity, machine learning models extract interaction information and their effects from textual sources. The 2013 DDIExtraction Challenge, organized by researchers at Carlos III University, established a standardized dataset to evaluate algorithms’ proficiency in detecting interacting drugs and characterizing their relationships; this dataset continues to serve as a key benchmark.
Additional computational methods analyze trends in user-generated data such as electronic health records and adverse event reports. Regulatory agencies like the FDA’s Adverse Event Reporting System (FAERS) and the WHO’s VigiBase collect reports on adverse drug reactions. Deep learning techniques process these datasets to identify potential drug-drug interactions, thereby advancing pharmacovigilance initiatives.