Welcome back to Healthy Innovations! 👋
This week's Deep Dive is about cardiac digital twins – personalized computer models of a patient's own heart that let doctors rehearse a procedure before touching the real thing. It's moved fast in 2026 alone: from a single hospital reporting a clinical trial to doctoral training programs and early commercial products, all in the same year rather than the usual decade-long lag between research and practice. That pace, more than any single result, is what made this one worth writing up.
Let's dive in!
Creating digital copies of the heart
Every year, thousands of heart attack survivors develop a dangerous irregular heartbeat called ventricular tachycardia. Doctors treat it with a procedure called catheter ablation: threading a thin tube into the heart and burning away the faulty tissue causing it.
That fix doesn't always work the first time.
Dean Ruther, a 64-year-old Iowa farmer, found that out the hard way: his first attempt failed, and he spent weeks afterward wondering, in his words, "what if they can't fix this?" A second attempt worked. Reported success rates for this procedure range from about 60 to 80%, based on historical comparisons, which still leaves many patients facing exactly what Ruther feared.
That gap is what a new wave of virtual hearts is designed to close: personalized copies of a patient's own heart, used to rehearse a procedure or test a device before anyone touches the real organ. It's moved fast this year alone. A Johns Hopkins trial published in April has already tested the idea in real patients, and King's College London launched a doctoral program in July to train the next generation of researchers who will build these models, one of several signs this is becoming a genuine medical field.
What a virtual heart actually is
Doctors and engineers call this a digital twin: a computer model of a patient's own heart, built from their MRI or CT scans, with healthy and damaged tissue mapped separately. Simulation software then predicts how electrical signals travel through that specific heart, including the faulty circuits that cause a dangerous rhythm, or how an implant will sit once placed. Aircraft manufacturers have used the same idea for decades, testing a virtual engine before touching the real one; cardiology is now applying that logic to a beating heart.
That matters because of how this procedure is normally done.
Doctors currently find the faulty circuit in real time, probing the heart and reading how it responds to electrical pulses as they go. Tissue that looks the same on the day can behave very differently once burned, so some of what gets treated turns out, afterward, to have been the wrong spot. That live guesswork is a meaningful part of why the procedure fails as often as it does. A model built in advance changes when that guessing happens: before the procedure, on a screen, rather than during it, inside a patient.
Inside the first trial
A Johns Hopkins team ran a small, single-hospital feasibility trial, published in the New England Journal of Medicine this April, testing the approach in ten patients with this irregular heartbeat. Cardiologist Jonathan Chrispin and biomedical engineer Natalia Trayanova built a virtual heart for each patient from MRI scans, then simulated its electrical activity with an AI tool the team has described elsewhere as DIMON, which cuts simulation time from hours to seconds. Doctors used that simulation to find the exact circuit causing the problem and rehearsed the procedure virtually before making a single real burn.
All ten procedures worked immediately – doctors could no longer trigger the abnormal rhythm once they finished – and eight of the ten patients had no return of it, off heart-rhythm medication, over an average follow-up of about 13 months. The other two had brief early recurrences during healing but were stabilized without a repeat procedure.
Image source: King’s College London
Training the workforce to build these hearts
A single trial only helps a handful of patients unless people exist to build and check these models more widely. King's College London launched CDTnet in July, a doctoral program training 15 PhD researchers specifically in this kind of heart modeling, recruiting from September. It builds on the university's existing DT4Health center and its Personalised In-Silico Cardiology initiative, funded through a European Union research fellowship.
A pipeline of people trained to build and check these models across different hospitals is what turns one trial into everyday care.
Companies moving this toward the clinic
The Hopkins study is a research trial, but pieces of the same science are already heading toward the clinic and the market, at different stages, on both sides of the Atlantic:
CardioSolv Ablation Technologies (Baltimore), a company building on the same published science as Trayanova's earlier work, is developing patient-specific simulation software to guide this kind of procedure, working toward regulatory clearance for smaller, more precise treatment.
FEops (Ghent, Belgium) makes HEARTguide, an FDA-cleared virtual-heart platform used to plan a different heart procedure: closing off a small pouch in the heart to lower stroke risk in some patients with atrial fibrillation. Doctors use it to test different implant sizes and positions before choosing what to place in a real patient.
Dassault Systèmes (France), better known for industrial engineering software, runs the Living Heart Project, a detailed virtual heart model built with hundreds of clinicians and industry partners. It's working with the FDA to explore whether these simulated hearts can help speed up how new cardiac devices get approved.
Between them, these efforts suggest virtual hearts are moving toward everyday medical practice, even if most are still early-stage.
Simulating whole populations
In Barcelona, Elem Biotech, a spin-out of the Barcelona Supercomputing Center, works at a different scale again: building populations of thousands of virtual hearts, rather than single-patient copies, to test new cardiac devices and drugs on simulated patient groups before human trials, reducing how many real patients and animals are exposed to unproven therapies.
Virtual twin hearts. Image source: Elem
Line up all five efforts and a pipeline emerges: Johns Hopkins and CardioSolv are focused on the single patient in the room, FEops has an FDA-cleared product already in use, Dassault Systèmes is working with regulators on testing standards, King's College London is training the people who'll build the next hundred versions, and Elem Biotech is testing what should be offered to the next million patients.
None of this is everyday care yet. Early results are encouraging. The harder question is whether the infrastructure now being built across Baltimore, Ghent, France, London and Barcelona can turn a ten-patient trial into ordinary hospital practice, so the next patient facing a first procedure gets a truer answer going in than Ruther did.
10x the context. Half the time.
Speak your prompts into ChatGPT or Claude and get detailed, paste-ready input that actually gives you useful output. Wispr Flow captures what you'd cut when typing. Free on Mac, Windows, and iPhone.
Innovation highlights
🩸 A $60 test that skips DNA sequencing. Researchers at Tel Aviv University built a blood test that spots lung cancer by reading DNA methylation patterns instead of sequencing it outright, cutting cost and turnaround to about $60 and two to three days. In a 103-person study, it hit 93.1% sensitivity and 90.3% specificity for stage 2 to 4 disease. It's early proof-of-concept, not yet validated for stage 1 or screening, but a methylation read is a genuinely different lever than sequencing, not a cheaper version of the same test.
🩹 The patch that redoses itself. Virginia Tech's iNal patch uses 121 microneedles to reach fluid under the skin, where silica nanoparticles loaded with naloxone sit behind fentanyl-sensitive molecular gates that open on contact with fentanyl. Because only some pores open each time, the patch keeps a reserve to respond again if fentanyl levels stay high, addressing the problem where naloxone wears off before fentanyl does. In lab tests it reacted within minutes and kept working for 24 hours; in mice, it responded across at least three separate exposures with no signs of irritation.
🎯 Tricking tumors into a target. Solid tumors have resisted CAR-T therapy for years because they lack a clean marker for engineered immune cells to grab onto. USC researchers built SHIFTERS, using focused ultrasound plus a tumor's low-oxygen environment as a "lock and two keys" to force cancer cells to display CD19, a marker CAR-T already recognizes. In animal models with human brain and liver tumors, treated tumors shrank while untreated ones kept growing, and just 10 to 25% of cells needed the marker to trigger a tumor-wide attack.
Cool tool
🧠 LumosityRx is a prescription digital therapeutic for adults 22 to 55 with inattentive or combined-type ADHD. Cleared by the FDA under the name Prismira, it delivers 13 adapted cognitive-training games – one is called "Trouble Brewing" – in a personalised, adaptive 15-minutes-a-day regimen over nine weeks, meant to sit alongside medication and clinician-directed therapy, not replace them. It's currently taking waitlist signups rather than shipping, and will be available in seven languages once live.
The evidence is the pivotal GAMES study: a 560-person, randomised, double-blind, sham-controlled trial. Users improved 1.1 points versus 0.3 on the TOVA attention test, 32.1% were rated "much improved" by blinded clinicians, and quality of life rose 8.7 points on a standard questionnaire, with only 0.7% reporting frustration and no serious adverse events. It won't necessarily ease hyperactivity, only measured attention – and unlike Lumosity's consumer brain-training games, which drew an FTC false-advertising settlement in 2016, this version carries FDA clearance and trial data behind a narrow, specific claim.
Image source: LumosityRx
Weird and wonderful
Hiroshima University researchers tracked 339 pancreatic cancer patients who had major surgery and found something odd: those with 21 or more of their natural teeth survived a median of 60.7 months afterward, versus 39.1 months for people with fewer – nearly two extra years. It held up even after accounting for tumor stage, chemotherapy, and nutrition. Oddly, how well people could actually chew didn't predict survival, only the raw tooth count did, ruling out "you can eat better" as the simple explanation.
Researchers suspect tooth loss over a lifetime tracks something bigger: chronic inflammation, frailty, and general wear and tear the body has quietly been logging for decades. As lead researcher Kenichiro Uemura put it, "the mouth may reflect the body's long-term resilience." It's one hospital and one patient population, so treat it as a hint rather than a verdict – but it's still a good excuse to finally book that overdue dentist appointment.
Image created using Canva AI
Thank you for reading the Healthy Innovations newsletter!
Keep an eye out for next week’s issue, where I will highlight the healthcare innovations you need to know about.
Have a great week!
Alison ✨
P.S. If you enjoyed reading the Healthy Innovations newsletter, please subscribe so I know the content is valuable to you!
