Patient selection tools that speed up clinical trials
By Perceiv AI | Published on March 14, 2022
Problem and objectives
Problem
- Getting a drug to market requires successful clinical trials, but clinical trials are long and expensive: 5 to 13 years and about US$700M for Phases 1, 2 and 3 (2014).
- 75% of clinical trials fail (difficult to predict patient health outcomes)
- Participation of patients not relevant to Alzheimer’s clinical trials costs US$4.3B
- Significant percentage of patients disqualified for clinical trials due to heterogeneous outcomes: 30% of patients with early-onset Alzheimer’s, 80% of patients with early Alzheimer’s symptoms, 90% of patients post-ACS (Acute Coronary Syndrome)
Objectives
- Predict the suitability of candidate patients for clinical trials (Alzheimer’s, cardiovascular disease)
- Make clinical trial data more homogeneous
- Facilitate successful clinical trials
Proposed solution
Solution
Perceiv AI is a Québec company operating in the health care field that develops patient selection tools to facilitate clinical trials. Its aim is to step up research on Alzheimer’s disease and heart disease. To do so, it has set up a prediction platform known as Foresight™.
The company’s key products are:
- Patient selection tool for Alzheimer’s clinical trials
- Patient selection tool for heart disease clinical trials
Principle
Foresight is a platform for predicting the short-term clinical course (prognosis). It is a tool for patient selection and enrichment to speed up clinical trials. In a longitudinal database with more than 550K patients, patients are selected based a prediction of changes in their state of health during the clinical trial. It targets chronic diseases: Alzheimer’s disease and cardiovascular disease.
Data
- Imaging
- Molecular
- Clinical
- Blood tests
- Genetic
User experience
The patient’s data is entered into Foresight™. The platform predicts the patient’s risk of progression of their disease. Depending on the outcome, the patient is included in the clinical trial or not.
The data from patients recruited into clinical trials using Foresight™ is enhanced and of higher quality, significantly reducing the clinical trial costs and time by up to 51% in some cases.
Foresight™ is compliant with HIPAA and ISO 27001 standards.
Role of AI
Algorithms that are used
- Machine learning
- Classification
- Statistical calculations
- Deep learning
- Convolutional neural networks
Impacts
Main outcomes
Experience and patient care:
- 40 to 60% of non-relevant patients excluded
- 20 to 50% of relevant patients additionally
Medical team wellbeing:
- Reduction in clinical trial time
- Reduced risk of clinical trial failure
- Roughly 30-50% reduction in the number of patients needed per clinical study
Public health:
Increase clinical trial quality
Health costs:
- Reduction in trial cost with control patient groups
- Reducing the risk of clinical trial failure
Generation and dissemination of new knowledge:
- Two scientific papers published
- Five scientific presentations (2021 Alzheimer's Association International Conference and 2021 Clinical Trials on Alzheimer's Disease Conference)
Economic development:
- Increase in staff from 2 to 5 employees between 2018 and 2021
- Target of +15 employees (scientists, developers and marketing staff) by the second quarter of 2022
Benefits in numbers:
- Without Perceiv AI: cohort not significantly enriched, 40% capacity to detect significant effects; cost of phase 3: CAD$287M
- With Perceiv AI: enriched high-quality cohort; 90% capacity to detect significant effects; cost of phase 3: CAD$219M
Paper published showing gain of one year and savings of CAD$68M
Challenges takled
It is difficult for startups to qualify for public funding because they must already have funds available to be eligible (public funds require contributions from the company up to 50% in many cases). Without revenue and seed money, startups cannot get a grant, so they cannot develop their products or services.
Conditions for success
Team mobilised
- 5 full-time employees (4 with PhD)
- 2 interns
Areas of expertise
- Data science and machine learning
- Health, precision medicine, pharmaceutical science
Advisory committee
5 advisors: Yoshua Bengio (MILA), Betsabeh Madani Hermann (Borealis Ventures), Manon Boisclair (Syantra), Serge Gauthier (McGill University), Robert Amyot (Medfar)
Collaborations
The company is working with hospitals to develop its product.
- CHUM
- Douglas Mental Health University Institute
- Montreal Heart Institute
- Institut universitaire de gériatrie de Montréal
The company is working with research centres to develop its product:
- MILA
- McGill University
- Université de Montréal
Perceiv AI benefited from incubation in technology accelerators for the company’s development.
- Centech
- CTS santé
- CLSI San Francisco
- District3
Funding received
- Personal funds: in kind (co-founders’ work not paid)
- Sales
- Public funding (2021):
- NRC: CAD$200K project
- CQDM, MEI: CAD$600K project
Project stages
Launch of Foresight™ for Alzheimer’s (second quarter of 2022)
Project stages - to come
Launch of Foresight™ for cardiovascular disease (end of 2022)
Cloud platform development and optimization (end of 2022)
Context of the solution
Customers
Perceiv AI works with pharmaceutical companies to help them optimize their clinical trials.
- Acasti Pharma Inc.
- Dalcor Pharmaceuticals
- VTV Therapeutics
Short-term business model – B2B
Funding for IP and database generation for pharmaceutical companies
Medium-term business model – SaaS
Cloud solution licence with predictive algorithms for research centres and clinical trials under development
Long-term business model
Reference database for clinical trials in research centres
Market opportunity
Launch of round of financing in 2021: objective of US$2M in investments