Successful clinical trials occur at a ratio of 1 in 10, costing around $2-$3 billion, with drugs taking 10–12 years to be approved.
High costs, 90% failure rate, and the lengthy timeline have all contributed to Pharma’s incompetence.
The technique of using multiple compounds in the trial phase to increase the chances of drug approval has only led to increasing costs.
Dubbed as Eroom’s law, the industry observes that over time and despite technological advancements, drug discovery has become slower and costlier.
Johns Hopkins Bloomberg School of Public Health researchers noted that the loss is larger for clinical trials that involve more patients and last longer.
Around $6 million was wasted in trials that failed at phases I and II, and a whopping $77 million in those that failed at phase III.
What if life sciences data and AI can be leveraged to predict clinical trial outcomes?
Life Science data is a vast ocean offering scientists and researchers valuable information about compounds and connections between biological entities, as well as offering new possibilities for leveraging AI in ways that could increase both the productivity and ROI for pharma and biotechs.
IDC’s report named ‘Data Age 2025’ observed a 61% compound annual rate of growth in data, leading to a DataSphere of 175ZB by 2025, from the known 33ZB in 2018.
For life science alone, Densen estimated that it will double in just about 11 weeks by next year.
This is supported by another prediction that medical data (patients’ X-rays, MRIs, CT scans) will grow 300% between 2017 and 2020.
With increasing health issues all over the world and the need for faster and cheaper drug development, it is necessary to start leveraging the vast data ocean of life sciences.
The clinical stage, being the most expensive for site selection, patient enrolling, and testing procedures require accurate decision making.
5,000 to 10,000 chemical compounds are tested in preclinical.
Of these, 250 will show promise for animal testing and only 10 qualify for the clinical stage.
However, what if you could tell whether the clinical trial you are conducting will be a success or not?
Of course, some may argue that if you could there wouldn’t be a need to conduct the trial in the first place.
False negatives are a common phenomenon and can easily be avoided with artificial intelligence programs.
Each year, a few new drugs reach the market, but behind those, a number of failed ones contribute to large spends.
Approximately 86% fail to meet the recruitment targets on time, 57% show limited efficacy, underpowered samples, and poor statistical endpoints.
The dropout rate of trials has been observed between 15 and 40%.
Applying AI techniques on biomedical, real-world, and outside trial data could overturn the failure rate of clinical trials.
Predicting the success of clinical trials could redirect the focus of pharma and biotech companies onto compounds with the most potential, leading to more success with less expenditure.
This would also shorten the timeline for drug approvals.
Using AI for Clinical Trial Prediction
By leveraging advanced deep learning techniques based on publicly available trial data, as well as on real-world events that are continuously crawled, aggregated, and analyzed, Innoplexus designed a CTP Engine to predict the outcome of clinical trials.
To make sense of the drug development landscape, we have developed a neural network that takes in drug compound characteristics, clinical trial features and sponsor track records.
The AI and advanced analytics-enabled engine is fully automated and calculates predictions for ongoing trials in real time, continuously accounting for new information which may impact the probability of a trial meeting its endpoints.
The engine leverages around 230+ features, such as indication, drug, trial design, targeted patient population, and sponsor-related information.
By introducing AI-based approaches in the design of clinical trials, we aim to present the optimal set of parameters to set up a clinical study.
“Each trial is assigned a probability of meeting its endpoints.
All of this is done on a fully automated, continuous, and real-time basis.
This is enabling us to calculate predictions for all ongoing trials – automatically accounting for new information that might impact the probability of a trial meeting its endpoints”, explains Dr. Gunjan Bhardwaj, Founder & CEO, Innoplexus AG.
CTP Engine serves various stakeholders keen to benefit from the prediction of clinical trials, such as CROs, pharma, and biotech companies.
It can be used to track clinical trial key performance indicators (KPIs), optimize recruitment strategies, and mitigate operational and financial risks.
A pharma company’s stock performance depends heavily on the success of clinical trials.
Having prior knowledge of approval probability can, therefore, help investors gain an edge by capitalizing on our state-of-the-art technology and life science data ocean to guide their investment strategy.
CTP empowers pharma to make accurate decisions and course corrections before obstacles delay a trial.
It supports consulting firms to guide through M&A decisions and creates high value for investors to reap previously unattainable benefits and opportunities.