GlobalData has announced the accuracy of its upgraded predictive analytics tool “Likelihood of Approval (LoA)” model, powered by Artificial Intelligence (AI).

The upgraded model delivers greater accuracy in assessing the probability of a drug with specific characteristics progressing to the next stage of clinical development, as well as its ultimate probability of the drug receiving market authorisation.

This enables stakeholders such as pharma companies, investment firms and consulting firms to leverage LoA for portfolio prioritisation, risk assessment, deal, evaluation, and improve strategic decision-making across the pharmaceutical development lifecycle.

LoA estimates the probability of a drug receiving regulatory approval in the US. The tool employs 108 specialised Machine Learning (ML) models using Random Forest algorithms across 18 therapy areas, trained on a comprehensive 24-year historical dataset, including extensive data on approved drugs, failed developments and current pipeline drugs.

The model predicts the Phase Transition Success Rate (PTSR) for each clinical development stage from phase-I to pre-registration. The LoA is calculated by compounding the PTSR at the current and following stages the drug is yet to progress through.

The LoA database provides a detailed and comprehensive analysis and phase transition success probabilities for drugs in development. It features two distinct datasets:

• Drug-Specific LoA:Drug success rate prediction by using ML models with close to 20 data attributes analysed ranging from drug, clinical trial, regulatory and company.
• Benchmark LoA: Historical success rate across therapy areas, indications, molecule types (modalities) and targets/mechanism of actions.

The updated drug-specific LoA model now leverages advanced ML to make predictions, marking a full shift to pure ML for improved accuracy. As a result of these upgrades, the model’s prediction accuracy has improved significantly, to 85.4 percent.

The enhanced LoA model delivers value across multiple stakeholders, including pahrmaceutica and biotrch companies, investors and financial analysts, business development and licensing teams, regulatoryb affairs professionals, and consulting and advisory firms. 

The LoA platform enables pharmaceutical companies to make evidence-based decisions by providing AI-driven predictions, reducing the need for time-consuming manual research and potential human biases; helps stakeholders make informed strategic choices by quantifying the risk of regulatory failure or success; assists companies in prioritising drug candidates with higher approval probabilities, improving R&D efficiency and ROI and enables early identification of high-risk programmes, allowing proactive mitigation strategies.

Revati Tatake, Global Head—Pharma Research, Analysis and Competitive Intelligence, GlobalData, said, “With the regulatory approval in the US being the most critical milestone in drug development, GlobalData’s enhanced LoA model empowers pharmaceutical companies to assess LoA of competing assets, determine market share and make informed strategic decisions.”