Sorry, you need to enable JavaScript to visit this website.

Smart Clinical Signal Detector


Smart Clinical Signal Detector

Region: Global


Pfizer clinical teams have near real-time clinical and operational treatment-blinded, coded data access via multiple clinical data electronic systems (e.g. Case Report Form (CRF) Electronic Data Capture, patient safety review tool, randomization-trial-supply-management system, PRO and ClinRO reporting tools, other non-CRF data, statistical tools). AI/Machine Learning technology platform may be able to take in and mine data from these systems to identify potential trends and patterns, including in safety and efficacy data of investigational medicines or vaccines, which could generate clinical hypotheses to drive earlier data driven study decision making towards delivering much needed medicines and vaccines as efficiently as possible.

Clinical Data Challenge

  • Aggregate patterns or trends in historical data sets for clinical hypothesis generation is currently supported by individual clinicians performing manual treatment-blinded clinical data review using numerous, unconnected systems.  
  • Early detection of important patterns or trends in study data may help clinical teams across a broad variety of therapeutic areas (e.g., vaccines, inflammatory bowel disease, hyperlipidemias, oncology) to adapt study conduct and/or study oversight, including for example to refine the population currently under study, flag a safety signal, or provide an early warning of lack of efficacy.

Solution Benefits

  • AI/Machine Learning- based technology platform that processes information from multiple data systems and assessment of co-variance of key/select data points may have the ability to surface important and clinically meaningful trends or patterns during study conduct. The development of this AI/ML-based technology platform, with a human in the loop, in turn, could help study clinicians drive better and earlier data driven clinical and operational decision making for clinical development in Pfizer studies.
  • The ability for study clinicians to review large and diverse volumes of data and efficiently identify data patterns.

Challenge Statement

Historically, the industry has been limited to manual processes for clinical data review, which are very time-consuming and laborious. Pfizer’s innovation challenge ‘Smart Clinical Signal Detector’ is a challenge to external organizations (“Challenge Applicants”) to:

    1. develop and demonstrate an AI/Machine Learning solution to identify trends and patterns in Pfizer studies and assess: positive and negative correlations between eCOA data (e.g. validated Patient Reported Outcomes or Clinician Reported Outcomes global assessments) and (i) major inclusion criteria; (ii) select demographic data; (iii) laboratory safety or efficacy data; and (iv) known safety concerns;
    2. positive and negative correlations between at least 3 critical data points identified by a Challenge Applicant, such as a machine-learning generated variable identified after learning on dataset, and (i) major inclusion criteria; (ii) select demographic data; (iii) laboratory safety or efficacy data; and (iv) known safety concerns.

The successful Challenge Applicant would be able to demonstrate (a) and (b) ab  ove to Pfizer by accessing specific Pfizer-prepared clinical trial protocol, therapeutic-area information, and clinical trial data in a Pfizer-provided secure hosting environment. Examples include:   

(i) observe change in distribution of blinded efficacy data over time in association with select blinded key co-variates,

(ii) identify demographic data positive and negative correlations in association with select blinded key efficacy/co-variates trends,

(iii) identify inclusion criteria positive and negative correlations in association with select blinded key efficacy/co-variates trends.

Selection Criteria

  • Experience and understanding of the challenge statement(s) and experience in healthcare and/or clinical trial data markets where this solution could be used
  • Prior experience with Machine Learning and NLP is a must
  • Solution must be compliant with all applicable legal, privacy, and regulatory requirements
  • Proposed solution that is fit-for-purpose as described in this Challenge, including a solution that is structured to detect differentiating patterns using treatment-blinded, data points relevant to clinical hypothesis generation
  • Differentiation of solution within the digital technology landscape
  • Solution must be scalable across protocols and therapeutic areas in clinical development programs
  • Cost-effectiveness (to scale and to use)
  • Capability of successful applicant to develop and demonstrate the solution for future (near-term) global use.


  • Registration deadline: Accepting Challenge Applicants through Jan 07, 2022
  • Challenge start date: Feb 14, 2022 (it is subject to legal and technical prerequisites for the challenge)
  • Challenge end date: May 06, 2022(may change depending on the challenge start date)

Contact Information

Questions? Contact [email protected]