Glioblastoma is one of the most common, aggressive and deadly forms of brain cancer. The prognosis — with median survival of 15 to 18 months after diagnosis — hasn’t changed in more than half a century.
A major challenge in treating glioblastoma is that tumors vary widely among patients and the cells within them aren’t uniform. Too often, selecting the right combination of invasive surgeries, radiation and chemotherapy amounts to high-stakes guesswork.
A “digital twin” of the brain tumor could help. Thanks to the leaps forward in artificial intelligence and mathematical modeling over the past decade, doctors can now create a virtual copy of the affected area in a specific patient using genetic information, medical records, blood tests and scans.
With a digital twin, doctors can detect abnormalities before physical symptoms emerge, project how tumors will develop, and determine which combination of surgeries and treatments will deliver the best outcomes. Early results have been promising.
Digital twins are at the forefront of a long-awaited era of precision medicine — an important evolution from the one-size-fits-all approach to disease prevention and treatment. Today the technology might still seem closer to science fiction than reality, but with the right support, investment and safeguards, this can change.
The concept of a digital twin isn’t new. Digital twins have been used for decades in industrial sectors such as aerospace engineering and more recently in car manufacturing. With blueprints and real-time data from sensors, engineers can approximate performance in extreme conditions — a jet engine in severe weather, for example — and predict maintenance needs. Though definitions are evolving, digital twins typically require a two-way flow of information: Real-world data from a physical object goes into a model, then the model forecasts outcomes for the real-world object.
The potential uses of digital twins in medicine are wide-ranging. Beyond cancer, digital twins are being used to treat potentially fatal heart arrhythmias and manage Type 1 diabetes. Last year, Blackstone Inc. deployed a digital-twin pilot to curb spending on pricey weight-loss drugs for employees; costs fell by half for 160 people enrolled for at least three months. Scientists also hope to create digital twins for clinical trials for rare diseases and to improve the diversity of patient pools. Simply shrinking the placebo group would speed up trials, significantly cut costs and — hopefully — bring treatments to market quicker.
By some estimates, the digital-twin health-care market could grow to more than $20 billion by 2028 from $1.6 billion today. But there are challenges. Unlike a jet engine, the human body doesn’t have a blueprint. And while doctors collect copious amounts of data from patients, few electronic records systems interconnect.
The first step toward making digital twins mainstream will be getting more, and better, data. Europe, for example, is progressing toward a standard health database. While replicating such efforts will be challenging in the US given the fragmentation of the market, medical engineers should be building digital twins with the goal of making them interoperable across different data systems.
The computing power needed to run such models will require investment. Public-private partnerships, in line with the European Commission’s program to foster collaboration between industry and academia, could help defray costs. In addition, the technology demands new skills, bridging engineering and medicine.Privacy concerns will arise. Existing data protections will need to be reexamined and modified. There’ll be questions around who owns or controls a digital twin, and how diagnoses and treatments using the technology will be priced and paid for. These myriad uncertainties shouldn’t be allowed to impede development.
