Northwell Health Using AI & Natural Language Processing to Find Clinical Trial Candidates

Mar 16, 2019 | AI in Clinical Research, Clinical Trials, EHR, Natural Language Processing, Provider Health Record

Brain Neuro AI


Northwell Health is New York state’s largest healthcare providers. A nonprofit integrated healthcare network, it employs more than 86,000 employees. The system is home to 23 hospitals and more than 700 outpatient facilities. The system includes the Donald and Barbara Zucker School of Medicine at Hofstra/Northwell, the Feinstein Institute for Medical Research, urgent care centers, kidney dialysis centers, acute inpatient rehabilitation, sub-acute rehabilitation and skilled-nursing facilities, a home care network, a hospice network, as well as other services.  The system was formally called the North Shore-Long beach Island Jewish Health System.

Northwell recently announced it has adopted the software technology from Clinithink Technology to accelerate patient identification for clinical trials as reported in Health Data Management. Northwell decided to evaluate artificial intelligence (AI)-based solutions to accelerate patient identification for clinical research.  Their provider services span access to 8 million people who are served by 23 hospitals with more than 30,000 clinicians and 700 outpatient facilities.

The Clinithink solution supports Northwell’s use of AI and natural language processing (NLP) to support the rapid identification and qualification of patients for clinical trials—it is anticipated that this process and new technology will help in cases where patients are previously missed by more manual searchers.

Proof of Concept

Health Data Management reported that Northwell sponsored a proof of concept (POC) with the following metrics:

Total patient cohort considered:            939,378 patients

Associated documents for review:         3.3 million

Clinithink assessed the patients against 22 different criteria in 30 hours finding 88 highly relevant patients for the specific trial studied in the POC.  Elaine Brennan managing director of pharmaceutical ventures at Northwell was quoted in the article “Clinithink allows us to mine a much richer data set and identify patients across our entire network, not just the specific location where the study is being conducted.”

Who is Clinithink?

Formed in 2009, the UK and U.S.-based venture was founded by Chris Tackaberry and Peter Johnson who spent tenure at NHS.  They appear to have about 25 to 50 employees.  CrunchBase reports 3 funding rounds.  Owler estimates $6 million in revenues.


CLIX-CNLP forms the basis of their solution.


CLIX-CNLP benefits include:

  • Access to and standardization of clinically rich unstructured narrative
  • Overcomes limitations of traditional NLP for clinical use
  • Enables use of unstructured data to support new revenue models based on data-driven patient outcomes
  • Broadens ability to capture mandated clinical data thereby reducing information gap


Key features include:

  • On-demand interpretation of unstructured data
  • Available in the cloud or on premise
  • API access via SOAP or REST
  • Specialty-specific optimization capability
  • Coded output refinement for clinical specific content
  • Map output to standardized terminology such as ICD, RxNorm, etc.

The system purports to enable the research organization to extract meaningful, computable information from unstructured healthcare narrative. Leveraging sophisticated clinical language models and algorithms, it understands and transforms unstructured data’s clinical meaning into rich standardized terminology. TrialSite New provides a link to a CLIX-CNLP white paper.  Documented case studies include:

Mount Sinai Retrospective Trial in Patient Group with Rare Kidney Disease

Mount Sinai Prospective Trial in Patient Group with Diabetic Nephropathy

AMBU Prospective Trial in Patient Group with Type 2 Diabetes Mellitus & other Specific Concomitant Disease

Patient Recruitment challenges represent one of the biggest bottlenecks for clinical trials.  Clinithink may be on the path to helping to solve at least some of the challenges.


Pin It on Pinterest