AI shapes the future of precision medicine
The explosion of data, increasingly powerful computer hardware and the availability of software frameworks are all enabling scientists to study and implement precision medicine initiatives that will allow treatments to be more accurately prescribed based on an individual’s genetic and molecular make-up, environment and lifestyle.
AstraZeneca, for example, has put significant effort into the field. Ruth March, Senior Vice-President, Precision Medicine and Biosamples, AstraZeneca, says: “Evolving technology is now allowing us to move from single genetic tests to a wide range of diagnostic approaches using rich scientific data from genetic and molecular biomarkers.
“Our teams are also developing tests that consider clinical and patient use right from the start of the drug discovery process. We have built global partnerships with emerging and established diagnostic companies, so that we can include the latest technological advances into our testing approaches,” says March.
“This is also beneficial for physicians and patients to select the best treatment path earlier in disease. We work closely with regulatory agencies and other healthcare institutions to determine how we can reduce access barriers across the globe.”
AstraZeneca launched its Precision Medicine Academy in November 2021to support the development of expertise and leaders in precision medicine. Targeted at early career researchers, the academy enrols cohorts from diverse locations and backgrounds. Over the three-year course, members have the opportunity to work with leading mentors from industry and academia, engage in theory learning and gain valuable practical experience working on drug-diagnostic projects.
“Not only are we developing talent in precision medicine, we are also pushing the boundaries of technology,” says March. “Collaborations with leading academic and biotech partners have driven advances in sequencing technologies where we can now sequence an entire genome in a matter of hours with a device small enough to fit in your pocket.
“This unprecedented progress is aligned to initiatives from industry and academia to bring diagnostic testing nearer to patients worldwide. Our aim is to make testing more convenient for patients, hence helping those eligible gain access to the best treatments for them.”
Making oncology treatment more precise
Researchers have developed a decision support system called GI TARGET (Gastrointestinal Treatment Assistance Regarding Genomic Evaluation of Tumors), which is being used to integrate precision medicine with oncology treatment. This may make it more widely available for the routine care of patients with gastrointestinal cancer, according to research published in JCO Precision Oncology.
The system provides guidance for the “interpretation and implementation” of tumour molecular profiling results. Researchers found that the design was both scalable and sustainable.
Precision oncology involves identifying the molecular changes that drive malignancy in an individual tumour. Targeting these oncogenic alterations is a strategy that can lead to different therapeutic approaches in patients with the same cancer diagnosis. Although achieving improved or enhanced clinical responses is the primary interest of this research, targeted therapies may also help alleviate some of the toxicities and morbidity associated with cytotoxic cancer treatment.
The research paper from Keller et al reports that: “With the growing number of available targeted therapeutics and molecular biomarkers, the optimal care of patients with cancer now depends on a comprehensive understanding of the rapidly evolving landscape of precision oncology, which can be challenging for oncologists to navigate alone.”
The researchers developed and implemented GI TARGET, a “precision oncology decision support system”, within the Gastrointestinal Cancer Center at the Dana-Farber Cancer Institute in Boston, USA. The multidisciplinary team systematically reviewed tumour molecular profiling for GI tumours and provided molecularly informed clinical recommendations, which included “identifying appropriate clinical trials aided by the computational matching platform MatchMiner, suggesting targeted therapy options on or off the US Food and Drug Administration-approved label, and consideration of additional or orthogonal molecular testing.”
The researchers tested the GI TARGET system using genomic data from 542 patients with gastrointestinal cancers who underwent tumour molecular profiling with OncoPanel or other tests in the first six months of 2019. Ultimately, the researchers developed a system for programmatic review of tumour molecular testing to inform clinical decision-making. The research paper states: “While the promise of precision oncology is yet to be fully realised, it is increasingly regarded as another tool of the trade. We have described in detail an example program for the systematic assessment of molecularly guided treatment and clinical care options for patients with cancer on the basis of tumour profiling.”
Sharing data to train AI models
While AI frameworks may be enabling precision medicine to see real-world applications for patients in early trials, it is the availability of patient data that underpins this entire paradigm of medical research. Patient data is much more highly regulated and harder to share than traditional laboratory data.
Adding these additional challenges to the more widespread issues with all laboratory data sharing, such as the interoperability of data created between multiple, disparate sources, means researchers can find it hard to use these new systems.
Collaboration and supporting frameworks that allow patient data to be shared for research purposes are opening up new avenues of research, providing researchers with sufficient patient records to enable them to build and train the initial models. Allina Health announced in early 2023 that it had joined the Guardian Research Network (GRN), a nationwide, non-profit, research healthcare network focused on advancing technology to accelerate cures for life-threatening diseases such as cancer.
The relationship is part of Allina Health’s plans to expand its research program to grow translational research, linking research with clinical activities. The new affiliation also contributes to GRN’s real-world data (RWD) that can potentially improve the health of millions of patients across the country. Translational research uses observations from trial science and patient studies to learn more about a disease, uncover powerful treatment alternatives, and move targeted therapy forward.
Badrinath Konety, President of AHCI and Chief Academic Officer at Allina Health, says: “Specifically, GRN is relevant for the Allina Health Cancer Institute (AHCI) and for other areas of non-cancer research, including cardiovascular and neuroscience. Cancer is primarily a disease of genetic abnormalities. GRN has the data and tissue for potential research that can drive precision medicine and help to recognise disease-causing variants.
“GRN helps us keep patients close to home by offering our patients a chance to participate in trials, including those complementing Allina Health’s Population Health strategy,” says Mike Koroscik, Vice-President, of Oncology at Allina Health. “GRN will accelerate needed precision medicine and translational trial access, broadening our research offerings. These novel clinical trials are usually only available in major academic medical centres, but GRN makes them accessible at all our AHCI sites in the region.”
GRN organises highly complex RWD to answer research questions in oncology, rare diseases, diabetes, and other genetic diseases nationwide. Research teams such as Allina Health’s can bring pharmaceutical and diagnostic development studies to patients who may benefit from early detection and new therapies.
“We’re excited to bring Allina Health into the GRN network,” says Shirley Trainor-Thomas, Vice-President Partner Network of GRN. “The oncology efforts of the AHCI combined with Allina Health’s research expertise in other areas is a testament to their commitment to improving patient outcomes, promoting health, and raising awareness about the importance of genetic data and diagnosis in healthcare. GRN and its other member health systems are committed to following the data and pushing science forward by finding new medical procedures, medications, and treatments based on aggregate RWD and data analysis that accelerates cures for diseases.
Entry into GRN’s network of health systems, with patients in more than 32 US states, offers Allina Health the potential for more patient treatment options and high-impact study participation. New projects, including those with Allina Health patients, are expected to drive healthcare benefits to patients across the US.
Looking to the future
Combining secure and robust data sources, integrating that data into your own medical data and applying the necessary models from AI or that have been specifically created for precision medicine will enable scientists to start developing treatments that are much more targeted. These can be based on a patient’s specific genetic characteristics or additional profiling created from other health or environmental metrics. Building competency and expertise in this burgeoning field will take some time, but that is not to say that organisations are starting in 2023 with no experience. “At AstraZeneca, we first started applying a precision medicine approach to our drug discovery over 10 years
ago,” says March.
AstraZeneca was “identifying and testing specific biomarkers that might guide the development of candidate drug molecules, starting with oncology. At the time, we focused on candidate medicines that were late in development and retrospectively analysed data from our clinical trials to identify patients who had responded to treatment,” she says.
“We soon started to look across the spectrum of cancers, identifying genes responsible for tumour growth and developed single gene tests alongside the process of drug development, so we could identify and enrol patients right at the start of clinical trials who were most likely to benefit from targeted therapy.
“Precision medicine is now being applied across 90% of our portfolio, we are taking our learnings and success from oncology and applying them to complex chronic diseases, like asthma or heart failure, using the 5R research framework (right target, right patient, right tissue, right safety, right commercial potential).
“With this expansion across our therapeutic areas comes new areas of science and technology. Using artificial intelligence, or AI, we can test up to a million biomarker models in early-phase clinical trials, which is something that would be impossible by humans alone. In addition, we are building digital apps, similar to those used during the Covid-19 pandemic, to simplify patients’ experience when they enrol for diagnostic testing in our clinical trials,” says March.
Case study: Transforming biopharma development
BIOVIA ONE Lab improves efficiency and simplifies the informatics landscape at a global leader in biotechnology therapeutics
Recently, a biotechnology company selected BIOVIA as a strategic partner to deliver and deploy ONE Lab across their process development organisation.
Challenge: Connecting data silos and streamlining workflows
With nearly 2,000 scientists and twice as many instruments spread across multiple global locations, the customer was managing eight different electronic notebook (ELN) systems and laboratory information management systems (LIMS). They faced a seemingly insurmountable task attempting to network and integrate such a large number of disparate systems. They also dealt with the constant risk of transcription errors because data was being manually transferred between systems, sometimes with spreadsheets as the intermediate step.
An immense amount of time was spent ensuring the validity and integrity of the data throughout the discovery and development process. Experimental data was further divided between small and large molecule divisions, with no way to cross-interrogate the information.
The customer’s challenge was to unify data collection and management across the different phases of early discovery, R&D, and clinical and commercial manufacturing. The scale and complexity of this challenge was enormous, completely changing the way data is handled throughout all aspects of their organisation. If the data identification could be unified, the data itself becomes smarter and self-aggregating, making it easier to locate and re-use, and more meaningful for the scientists.
Solution: A holistic lab environment with BIOVIA ONE Lab
The customer’s goal was to build a flexible and interconnected system, tie different systems and components together, and make them collaborative.
This holistic approach wouldremove complexity and streamline scientists’ daily workflows. By properly parameterising and standardising experimental, instrumental, and process properties in the ONE Lab solution, new laboratory processes are easily created as if working with building blocks.
Scientists can perform these tasks without the need for a software developer. This way of thinking extends throughout the system in multiple applications. When a new experiment is initiated, most parameters are prepopulated based on the chosen experiment, minimising data input by the scientist. The customer employs a comprehensive data lake to store and index all experimental results and metadata. ONE Lab feeds the data lake along with other systems, and maintains a contextualised index of interconnected information. Every piece of equipment is managed by the system, and each piece of data and metadata can now be accurately searched. As part of the ONE Lab solution, BIOVIA also helps the customer manage the delivery of results to scientists, delivering email notifications but not data; thus the data is not divested from the system to an inbox.
Results: Increased workflow efficiency, better data quality, and improved decision-making
With the majority of data entry and transcription now done automatically, the customer no longer needs to expend the same effort validating the integrity of their experimental data. Additionally, scientists no longer need to utilise multiple electronic systems to input and manage data throughout their experiments. The customer defined their data taxonomy based on the requirements for FDA submissions, meaning that the appropriate data is naturally aggregated, saving time in submission preparations. However, the data itself is also smarter – a sample in the system identifies what processes it has already undergone, and what are the appropriate next steps. Scientists can find previous related results, including those which would have previously been hidden as “dark data,” resulting in a 60% increase in scientific and engineering analysis efficiency.
The goal of an experiment is usually single in nature – to validate or invalidate a hypothesis. However, the future uses of an experiment’s data are theoretically infinite. By consistently capturing, linking, and referencing everything in the data lake, this knowledge is preserved for maximum impact. The customer also feeds this information into the decision layer of their system, which involves modelling, predictions and trend analysis to aid their business decisions. By consolidating the legacy LIMS and ELNs
into a single deployment of BIOVIA ONE Lab, the customer was able to greatly increase the efficiency of its process development organisation. The project capacity increased from 30 to nearly 100 over the course of two years, with no increase in headcount.
The efficiencies gained with ONE Lab resulted in more than $50m savings in operating expenses.