Predict success. With AI.
MH Predict is an application designed to predict clinical trial success using the latest developments in Artificial Intelligence (AI) and Machine Learning (ML).
Transparency is the foundation of trust in AI/ML algorithms. Users need to know what data the algorithms consider, and what thresholds are used in arriving at the conclusions, so that they can understand the results and reap the benefits AI offers with confidence. This is especially true for applications used in a medical context.
MH Predict creates trust by providing customers access to AI/ML algorithms in an easy and intuitive way.
MH Predict is an AI software that draws curated biomedical data from numerous sources to:
Predicting clinical trial success with MH Predict has the following benefits:
MH Predict’s value to R&D and commercial Pharma functions
Would you like to see a demo?
MH Predict clinical trial case
- Trial point prediction (success/failure)
- Probability of technical success
- Reliability graph
- Contextualization: compare the prediction to similar trials
- Feature effect simulation (in the UI, more detailed analysis offered ex-UI)
- Feature importance analysis – opening the black box by identifying key parameters influencing the prediction
Probability of trial success predicted by MH Predict relates to the real-world outcomes of clinical trials.
Drug That Celgene Bought For $710 Million Fails
Novartis heart drug portfolio hit by failed serelaxin study
Merck’s Keytruda fails late-stage study in liver cancer patients
Opdivo Fails to Extend Survival in SCLC Patients in Phase 3 Trial, Bristol-Myers Squibbs Reports
AbbVie takes $4B hit on Rova-T failures
OncoMed’s small cell lung cancer drug tarextumab fails in phase 2 trial
The disease: Atrial fibrillation
The developer: Bristol-Myers
Squibb, Pfizer Peak sales potential: Estimates start around $3.8 billion and rise to $5 billion or more.
AbbVie rheumatoid arthritis drug succeeds in late stage trial
Recon: Sanofi and Regeneron Find Success in Major Cholesterol Trial
Avelumab Becomes First Approved Treatment for Patients with Merkel Cell Carcinoma
AstraZeneca’s Lynparza shown to put brakes on ovarian cancer
Software as a service - intuitive user interface
Drug discovery and development are long, costly, and inefficient. The average costs of bringing a drug to market have risen to US$2.6bn, according to the latest analysis by DiMasi and colleagues of Tufts University (DiMasi JA, Grabowski HG, Hansen RW. Innovation in the pharmaceutical industry: new estimates of R&D costs. Journal of Health Economics 2016; 47:20-33.), mostly owing to an attrition rate of 11 failed projects for every successful one. AI now offers opportunities to improve this process at virtually every stage.
ML is a well-defined term that describes a number of computational methods (algorithms) that allow a computer program to learn from data without specific instructions and improve learning with more data (“experience”). Computer scientist Tom Mitchell came up with the most famous and clear-cut definition: “A computer program is said to learn from experience E with respect to some class of tasks T and performance measure P if its performance at tasks in T, as measured by P, improves with experience E.” ML is therefore a purely technical term. In contrast, AI is not as clearly defined. AI pertains to a concept rather than a method. There are points of contact with psychology, computer science, philosophy, biology and neuroscience. AI uses ML as a method, but solutions can also be created by non-ML systems, for example, knowledge systems. AI integrates ML into solutions. An autonomous car is an example of AI, with one component being ML algorithms. ML is thus a subset of AI. However, ML as a method has been the biggest contributor to the growing success of AI in recent years. In this sense, MH Predict can be referred to as both ML and AI.
ML applications are now being tried in virtually all steps of the drug discovery/development cycle and post-marketing to improve drug adherence. According to the Financial Times, in 2018 at least 15 big pharma and biotech companies used AI in parts of their drug development cycle. Most examples of the use of AI in pharma come from early drug discovery (target discovery, lead compound identification) or lead optimization, fields that have previously used quantitative approaches and statistical modelling. Progress in pattern recognition in pathological or radiological images and sensor signals opens up new opportunities for biomarker discovery for use in clinical trials (e.g., patient inclusion or endpoint definition).
AI is now also moving into the area of clinical operations with systems that optimize patient recruitment and site selection. MH Predict sits at the clinical stage of the drug development path but extends into the earlier translational space and strategic and business aspects within pharma.
Examples of usage and AI/ML deals in the drug development process include:
- Target discovery
- Lead compound identification
- Lead optimization/medicinal chemistry
- Patient selection
- Prediction of clinical success
- Clinical operations