Unlearn Secures $50M Series C to Advance AI-Powered Digital Twins for Accelerated Clinical Trials

Feb 29, 2024 15:00 CST Updated 15:00

Unlearn.AI

AI Drug Trial Instrument Developer

Due to recent changes in the official account push mechanism, we welcome you to set Boom Health as a "starred" account on the official account homepage to get the latest trends and expert insights in the digital healthcare industry第一时间!

On February 9, AI digital twin platform Unlearn.AI secured $50 million in Series C funding, led by Altimeter Capital. The company plans to use these funds to advance artificial intelligence by investing in its personnel, data, technical capabilities, and long-term R&D initiatives, eliminating trial and error in the medical field.

This article is approximately 1,500 words and will take about 3 minutes to read.

Unlearn.AI, Inc. was founded in 2017 and is headquartered in California, USA. The company brings together a team of experts from the pharmaceutical, medical technology, machine learning, and business fields. It is committed to leveraging machine learning and biostatistics to improve clinical trials, helping researchers achieve clinical trial goals with fewer patients. By analyzing patient health records, Unlearn creates digital twin models for each participant in clinical trials to predict changes in patient health under different treatments. This allows more patients to receive potentially beneficial experimental treatments, facilitates faster recruitment of participants and trial progress, thereby accelerating the time to market for new drugs.

Currently, Unlearn has completed 7 rounds of financing totaling $135 million:

Unlearn.AI Invents Twin RCT Solution, Enabling Biopharmaceutical Companies to Conduct Clinical Trials Faster and Improve Success Rates. Similar to traditional randomized controlled trials (RCTs), patients are randomly assigned to either a treatment group or a control group. However, unlike conventional trials, this solution utilizes machine learning models based on historical data to create prognostic digital twin models for each patient. After adjusting the prognostic scores of each patient’s digital twin model, the treatment effects on primary and secondary outcomes can be estimated more accurately.

Twin RCT Solution includes three aspects: First, aggregate historical data from previous trials to construct a large dataset of longitudinal patient data at the individual level; Second, utilize machine learning techniques based on artificial neural networks to build complex models from these extensive datasets; Finally, develop randomized controlled trials based on biostatistics of RCT design, using predictions from digital twin models to enhance efficiency and power.

Historical Control Data: Through machine learning from historical control group data, obtain user-customized standardized datasets in two steps:

Collect and Extract Clinical DataUnlearn selects specific diseases and collects historical data from relevant clinical trials;
Standardized Data: Standardize the extracted historical data according to specific requirements and transcribe it into the computer to complete the machine learning component.

Deep learning: Train the data filtered through Historical Control Data by setting corresponding conditions to obtain the target digital twin model.

The standardized dataset is divided into two groups (a training group and a test group), one for training proprietary machine learning models and the other for evaluation.
Train the model using the training dataset, which allows the model to learn key clinical trial points through the dataset.
After the model training is completed, use the test dataset to evaluate the model quality. Once it passes the inspection, the digital twin model is ready.

Ditigal Twin: After completing the above steps, introduce real patient data to generate a digital twin model of the patient.

Patients enroll in clinical trials and visit the site to collect baseline data, which will be used as experimental data to construct digital twins through the aforementioned model.
Clinical coordinators continuously follow up and collect patient trial data, and at the end of the trial, the clinical trajectory of the digital twin can be compared with the actual patient records.

This constructs a complete closed-loop clinical trial process, which is the general workflow of Unlearn.AI's digital twin solution, Twin RCT. This innovative method uses AI models trained on historical clinical data to generate predictive clinical records at the time of a patient’s initial registration of physiological indicators.

In an ideal scenario, the application of data twins can significantly reduce the number of patients recruited for control group trials by up to half, thereby greatly accelerating the trial process. This technological innovation not only enhances the efficiency and accuracy of clinical trials but also brings revolutionary changes to the development of the medical field.

Currently, the diseases involved by Unlearn include neurological disorders (AD, ALS, Parkinson's, etc.), immune-related diseases (Crohn’s disease), and cardio-metabolic conditions (stroke, diabetes, dyslipidemia, etc.).

In 2019, the company collaborated with the Critical Path Institute in the field of Alzheimer's disease. In February 2022, Unlearn.AI announced a multi-year partnership with Merck KGaA, located in Darmstadt, Germany. In 2023, QurAlis partnered with Unlearn.AI to accelerate and optimize ALS clinical trials using generative artificial intelligence technology. Through continuous collaborations with global biopharmaceutical companies, the company not only demonstrates the strength of its technology but also advances the further development of its technological platform.

Author: Zhang Qiankun

Share

Like

Tap to view