Machine Learning (ML) is a branch of Artificial Intelligence (AI) that allows computer systems to learn directly from examples, data, and experience. Enabling computers to perform specific tasks intelligently, ML can carry out complex processes by learning from data, rather than following pre-programmed rules.
In the world of clinical trials, mobile devices and real world data (RWD) sources can dramatically increase the amount and complexity of data collected, which can be difficult to manage. However, ML can be used to improve the quality, reliability, and availability of data by utilizing historical information to train ML algorithms and provide automated insights.
This post provides a quick overview of non-clinical and clinical examples of how ML is transforming the production and consumption of data.
First, let's look at some non-clinical ML innovations:
As the largest and oldest single-payer healthcare system in the world, the United Kingdom National Health Service (NHS) has identified huge potential savings by using analytics tools to make the most of its data. By optimizing patient treatment while reducing the use of less-effective medical procedures, it has recognized potential savings of over GB£ 1 Billion. By returning accurate, reliable data to clinicians and policy makers, it has also enabled antibiotic prescribing to be reduced by seven percent (7%).
“The NHS sits on billions of data points that have the potential to deliver tremendous value to the wider healthcare system in the UK, when combined and analyzed effectively. We can now do so much more with our data, resulting in significant savings for the NHS as a whole,” said Nina Monckton, Chief Insight Officer, NHS Business Services Authority.
Turning Point is a London-based, mental health charity working with healthcare providers to allow faster access to treatment. By using an online system that bypasses the UK’s referral process, wait times for medical care have dropped from four weeks to as little as two days. The charity employs ML enabled chatbots, offering access to help anywhere, anytime, to provide on-the-go advice. For example, someone who feels an anxiety attack coming on while stuck in a meeting can have a text conversation on his or her smartphone with a chatbot that provides advice on relaxation techniques.
Established in 1954, the European Organization for Nuclear Research, known as CERN, is the largest particle-physics laboratory in the world. CERN uses big data, cloud computing, and analytics to help researchers unravel the mysteries of the universe. Its Large Hadron Collider control systems produce vast quantities of systems-monitoring information. The CERN team is building ML models using the open source statistical and visual language R to predict potential failures and reduce maintenance.
Now, let’s look at some examples of ML and AI in safety, healthcare, and clinical use.
For pharmacovigilance, more than 60 percent (60%) of drug safety experts plan to use AI to improve the speed and security of adverse event case processing.
“Fortunately, adverse event processing is becoming faster and smarter with the help of AI and the cloud. Both technologies are helping drug safety experts to improve quality and accuracy in the handling of the data they work with and drive down their reporting costs,“ said Bruce Palsulich, VP of Safety Product Strategy, Oracle Health Sciences.
The likelihood of heart disease can be classified into five distinct values: absent, less likely, likely, highly likely, and present. A multi-classification ML and training dataset can be used to create a model that will predict the likelihood of heart disease in other patient populations. The project, training data and ML model can be downloaded from the Oracle Analytics Library.
Clinical data scientists have trained and applied ML workflows over transformation metadata to improve the quality of mappings. A team of data scientists used the Oracle Analytics Cloud (OAC) to explore ML training and application workflows to analyze the quality and consistency of historical, study-data-tabulation-model (SDTM) mapping. These approaches can improve the quality and understanding of clinical trial data, and identify trends that analysts could have missed.
The Potential for ML Innovation in Clinical Use
It’s clear that the process of evaluating large amounts of data from multiple sources for insight could overwhelm even large teams of skilled data scientists. In clinical research, ML-related approaches can help the industry to analyze the ever-increasing complexity of today’s clinical trials, manage continuous data collection from emerging data sources such as mobile devices, streamline safety case processing tasks, and verify consistency across data transformations. As the industry embraces these new technology approaches, there will be even more use cases that can benefit from AI.
If you’d like to learn more about ML, AI, and clinical data, contact firstname.lastname@example.org