Welcome Learning Health Systems!

After 10+ years of R&D and close to six years of live deployments across 20+ hospitals, today we proudly introduce Cognome to the rest of the world. Cognome’s mission is to bring the ground breaking work of Dr. Parsa Mirhaji and his team of data scientists, developers and clinical practitioners in the field of health data innovation, Artificial Intelligence and Machine Learning, to the rest of the world.

Cognome provides a suite of healthcare-specific ML models and LLMs, embedded within clinical and operational workflows, through an AI ecosystem built for tomorrow’s world of healthcare. Though few health systems have deployed ML models to this point, we predict in coming years ML models will explode across the healthcare landscape.?

Why? Because ML models deliver significant ROI by predicting clinical risk factors, intelligently automating time and resource consuming tasks, and because they learn and get smarter over time.

• Built with Intel, our Sepsis Prediction model is the best performing live model in the industry with a clinically validated 2% false-positive rate. Given its importance to patient safety, and because its personal to many of us, we are happy to share how we designed this high-performing model with our community of healthcare innovators.
• Our Case Cancellation Prevention model, one of the longest running ML models in healthcare, can save a large health system $500-$1M annually along with an improved experience for clinicians, staff and patients.
• Our chart abstraction LLM, AUTOCHART, is built on a 8B parameter model which fully automates the work of nursing abstraction, CPT code generation and data entry for orders, claims and registry submissions.

These are just a few examples of our high-value models and we have more in production, some of which are proprietary and others being built with partners and collaborators.

Currently, most ML models are managed individually likely at the code level. Cognome has built a market-leading, open Business Intelligence (BI) platform called TUNER BI. Our BI platform drills down into predictive risk factors, algorithm configuration parameters and performance analytics in real-time at the patient and population level. TUNER BI provides a single pane-of-glass view into your portfolio of AI / ML models and is extendable to Power BI, Tableau and other leading platforms.

To ensure transparency and proper governance, we designed TUNER BI on the four pillars of NIH’s AI Governance Framework: trustworthiness, fairness, transparency and accountability. TUNER BI delivers risk scores, predictions, trending and actionable analytics but also details bias, gender, race, ethnicity and many other components of a model trained on live patient data. This level of granularity and transparency has been a huge factor in fostering clinician trust and model adoption.

We are committed to transforming healthcare by collaborating and partnering with our community of mission-aligned innovators. James Green , Cognome’s CEO, summarized this quite well in his first blog describing AI-enabled Learning Health Systems, “Eventually, ML models will be everywhere serving side-by-side with clinicians as the primary form of clinical and operational decision support. This is one of the key capabilities that will transform hospitals into Learning Healthcare Systems. Together they will learn from one another and bring cohesiveness where there is otherwise entropy.” Our passion comes from the heart because we have seen the power of AI and ML to actually save lives and make healthcare better for everyone.

The AI revolution in healthcare is indeed here! We look forward to working with our community of innovators to make AI truly transformative and hugely impactful.

If the work we're doing is of interest to you, please follow Cognome on Linked In and contact us to keep updated on our progress.

Sincerely,

The Cognome Management and Advisory Team?