Jake Been, co-CEO of Neurophet, remembers the winter of 2022. Cash was thinning, investors were circling but not committing, and a disease-modifying therapy still felt like a mirage. “If an Alzheimer’s drug had not arrived, Neurophet might have shut down,” he said in an interview with Korea Biomedical Review (KBR). 

Betting the company on imaging software for a therapy that might never reach market was, as he now puts it, “either visionary or reckless.”

In 2023, Eisai and Biogen’s Leqembi (lecanemab) won FDA accelerated approval, and a new problem emerged for hospitals and drugmakers alike: how to measure whether the drugs were working, and how to keep patients safe.

That is where Neurophet says it fits. The Korean firm’s flagship platform, AQUA AD, is built to quantify amyloid clearance and detect ARIA, short for amyloid-related imaging abnormalities, the brain swelling or microbleeds that can force doctors to pause treatment. 

“Even a 0.2 cm difference in lesion size can determine whether treatment continues,” Been said. That level of precision is not something clinicians can reliably judge by eye, and drugmakers suddenly needed hard answers.

Now Eli Lilly and Roche are handing over guarded clinical trial datasets to test Neurophet’s claims. Those evaluations, Been said, represent a “make-or-break” moment. If the algorithms hold up, Neurophet could shift from niche diagnostic vendor to a standard part of Alzheimer’s treatment decisions.

A near miss

Neurophet was founded in 2016, just as late-stage Alzheimer’s candidates were still striking out and Lilly’s solanezumab added another miss that November. Even calling it an AI company felt off at the time. “We were really a brain science company,” Been said. “AI was a tool, not the purpose.”

The pivot came slowly. Neurophet started on the treatment side, building brain-modeling software to plan transcranial stimulation, and only later packaged its imaging work into diagnostic products. But Been kept circling back to Alzheimer’s, convinced that once therapies arrived, monitoring would become indispensable. 

Investors weren’t so sure. Series C fundraising stalled in late 2022, with potential backers questioning whether a market would ever materialize.

Then, in January 2023, Leqembi won accelerated FDA approval. By March, “Big Pharma started calling,” Been said, and the fundraising that had stalled restarted and closed by fall.

More important for the business than the cash, pharmas began allowing performance verification under data-transfer agreements, opening guarded trial data for Neurophet’s algorithms.

At the Radiological Society of North America meeting (RSNA) in December 2024, Neurophet was still talking about MS and early diagnostic clearances. By July 11, at a pre-IPO press conference, executives reframed the company’s story as completing pharma’s workflow. 

And in our August interview, Been dispensed with any hedging: “We’ve never really been just a diagnostic company. Our software is built for treatment -- for deciding who stays on therapy and who doesn’t." 

That narrowness was also a liability. Had Leqembi failed, Neurophet’s future might have failed with it. Been admits as much. “It was high risk, high reward. If no therapy ever came, we were finished. But now the therapies are here, and suddenly our pipeline is in demand.”

Neurophet’s decision to double down on Alzheimer’s rather than spread across multiple disease areas reflects lessons from peers. 

The story echos that of Lunit, another Korean AI firm that made its name in oncology. Been admits Neurophet studied that playbook. “They weren’t just thinking about radiology clients,” he said. “They were thinking about pharma.” 

Lunit went public on Kosdaq in July 2022 and then leaned into drug-facing work, co-developing pathology AI with AstraZeneca to triage EGFR-mutant NSCLC, a lung cancer subtype driven by EGFR mutations, from routine H&E slides, and has plugged its SCOPE tools into Roche’s navify platform. Alzheimer’s gave Neurophet no such anchor at the start.

That risk sharpened Neurophet’s approach. Where competitors often touted large portfolios, Been leaned into focus. “We saw others highlight having 10, 15 products,” he said. “But we believed concentration mattered more.”

The strategy shows in how the company talks about biomarkers. Neurophet didn’t try to cover every neurological condition; instead, it built tools around the specific markers tied to Alzheimer’s disease, then expanded into vascular dementia and MS as side extensions. 

That choice is now paying off, Been said, because hospitals need not just screening tools but confirmation and monitoring across the entire treatment journey. “If you don’t have confirmation solutions, you drop out. If you don’t have ARIA monitoring, you drop out,” he said. “We made sure we had both.”

Asia takes center stage

Geography has also worked in Neurophet’s favor. As with Kisunla (donanemab), the path ran U.S. first, then Japan, with the U.K. before China, and Korea not yet cleared.

Instead, Lilly is running an eight-country phase 3 bridging study, TRAILBLAZER-ALZ 5, recruiting at about 16 sites in Korea and targeting a first completion around July 2028, according to ClinicalTrials.gov. 

Lilly has not said when it will file locally, but Been said for Neurophet, Asia has shifted from afterthought to priority: prescriptions in Japan and China are “almost on par” with the U.S., and revenue across the countries is “nearly comparable.” “That is why we pushed so actively into Japan.”

Collecting imaging data is cheaper and faster in Korea. An amyloid PET scan typically runs about 1.0–1.5 million won here (about $730–$1,100). In Europe, self-pay prices generally land around €1,000–€5,000 or more (about $1,100–$5,500+), and in the U.S. a self-pay amyloid PET is often $3,000–$5,000 or higher, according to multiple industry estimates. 

Neurophet says its cost base and PET-first focus give it an edge over Western challengers like Cortechs.ai and icometrix, which emphasize MRI and keep PET quant tools in research use.

Cortechs.ai said in June it was preparing a 510(k) for NeuroQuant PET, and icometrix has been building MRI-led Alzheimer’s tools with Philips.

Still, finances loom. Neurophet posted a 14.6 billion won operating loss in 2024, with cumulative deficits above 60 billion won. Management projects break-even by 2027, but only if pharma partnerships turn into real contracts. 

Treatment’s second act

The bigger gamble is what happens once the current wave of anti-amyloid drugs is eclipsed. Kisunla already carries an FDA-cleared dosing update aimed at ARIA risk. Next up, Roche’s trontinemab is moving to phase 3 after early data showing rapid plaque clearance with low ARIA, and Lilly’s remternetug is already in phase 3. The field still remembers gantenerumab’s 2022 miss, so optimism is measured. 

Neurophet is betting that no drug will be perfect and that each new mechanism will need its own monitoring tools. “Our business model depends on there being the next generation,” Been said. “A single drug saturates fast. What we want is new mechanisms, new biomarkers, new reasons for hospitals to need us.”

That logic reframes pharma partnerships: the goal isn’t just to be bundled with Kisunla or Leqembi, but to be at the table for trial design of the next wave. Neurophet is already analyzing trial data for pharma, though Been declined to specify which companies beyond Lilly and Roche. 

Neurophet’s public story, then, has moved in stages: from FDA clearances for niche imaging tools, to IPO roadshows touting Alzheimer’s monitoring, to guarded pharma data evaluations. The company has learned to match its messaging to the moment, but the throughline -- a bet on Alzheimer’s becoming treatable -- hasn’t changed.

“We had no answer for years when people asked what happens if no therapy comes,” Been said. “Now the question is: who measures these drugs best?”