Artificial intelligence is being leveraged to solve a variety of healthcare challenges, but its greatest promise remains its potential to detect and diagnose diseases better than humans can.

The latter is being attempted by NeuraLight, an Israeli startup that is aiming to use computer vision and deep learning to analyze videos of a person’s face and eyes to determine whether the person is suffering from a neurological condition. A mere six months after launch, the company announced that it has raised $25 million last week. Koch Disruptive Technologies led the Series A funding round, with Breyer Capital, Samsung Next, VSC Ventures, as well individual investors. Including the money raised last week, the company has raised $30.5 million since its inception six months ago in late 2021.

So how does the technology work. Proprietary algorithms are used to extract information from videos taken by standard smartphones and webcams, explained Micha Breakstone, CEO of the company, in an email response to questions. Specifically, the AI is focused on oculometric data — data on tiny eye movements — which have been “demonstrated to be deeply correlated to the progression of neurodegenerative diseases in over 750 published scientific papers,” he said.

NeuraLight’s deep learning technology then” extracts over 100 of these ocular parameters (such as saccadic and anti-saccadic delay, how quickly the pupil dilates, and the eye’s blink rate) in one fell swoop to establish a robust proxy for the neurological health of a patient,” Breakstone said.

The short term goal for the company is to improve the design of neurological clinical trials and help to usher in better therapeutics for complex neurological conditions like Parkinson’s, Alzheimer’s Disease, ALS (Amyotrophic Lateral Sclerosis) and MS (Multiple Sclerosis). But Breakstone has an even more ambitious goal though he acknowledges that is years away — NeuraLight’s technology becomes the gold standard for neurological evaluation.

So that automatically means, current modes for neurological evaluation needs to be revisited. The main problem seems to be that disease detection and diagnosis is highly subjective, Breakstone believes, and has a 25%-30% inter rater-variability rate.

“A 25%-30% inter-rater variability in the evaluation of Parkinson’s basically means that if two doctors assess the same patient on the same day for Parkinson’s, the doctors present significantly different evaluations 25%-30% of the time,” Breakstone said.”This compromises treatment and presents a massive barrier for drug developers who can’t objectively measure whether or not, and the degree to which, a potential treatment for a disease like Parkinson’s (or MS or Alzheimer’s) is impacting the progression of the patient’s disease.”

And that variability, and lack of objective and sensitive endpoints has resulted in one thing, according to a news release from the company: a “dismal 6% of approval rate for neurological therapeutics, less than 50% the approval rate for non-neurological drugs.”

That variability and resulting scarcity of approved neurological drugs, is what can be potentially addresses with robust deep learning algorithms from NeuraLight. It is also what pharma can leverage to improve the design of neurological clinical trials in the future.

“The goal for pharma is a critical stepping stone to achieve our end goal. Namely, introduce our objective and sensitive ocular biomarkers as part of our decision-making platform, to increase accuracy in clinical trials and eliminate the subjectivity in neurological diagnosis,” Breakstone said.

He added that given that NeuraLight’s technology is device agnostic, researchers can incorporate the company’s technologies to conduct decentralized clinical trials, a move that is liberating clinical research from a central location and therby opening up the possibility of having a diverse trial population.

“[Being device agnostic] is particularly valuable for clinical trials that focus on neurodegenerative diseases that make it hard for those afflicted to travel to a central research facility,” he said.

While establishing NeuraLight-developed digital biomarkers as the gold standard in neurological evaluation is a few years down the road, here’s what’s possible now.

“In the immediate future we will be primarily contracted by pharma companies and will charge per project based on cohort size and label cadence,” he said. “This model will likely evolve as our biomarkers gain further clinical validation, and we become part of the indication itself and/or accepted outcome measure.

Pharmaceutical companies can use the company’s platform to design clinical trials and for that, NeuraLight will not need FDA’s blessing. However, in order to help pharma companies leverage its technology and develop new drugs, the company will pursue surrogate endpoints approval from the FDA in the future.

Now, the company will use some of the money raised to hire more people in addition to the 25 it already has. The rest of the money will help to build out NeuraLight’s own infrastructure and expand trials with research institutions and pharma partners. The company expects to get paid for its platform when it partners with pharma companies depending on the trial size and number of cohorts.

While NeuraLight is basking in its recent fundraise with a promise to fundamentally alter how neurological conditions are diagnosed and how drugs can be developed, it is not alone in its endeavor to develop alternative efforts to assess neurological disorders.

They include Altoida, which leverages AI and augmented reality to perform 10-minute cognitive assessments using smartphones from anywhere. Meanwhile, Beacon Biosignals is useing its EEG analytics platform to develop and validate neurobiomarkers to improve cohort selection during clinical trials, improve clinical assessments as well as identify novel endpoints.

For its part, NeuraLight is likely hoping that both its platform and its scientific advisory team can be a differentiator. In the news release announcing the fundraise last week, the company touted the fact that two Nobel Laureates — Professor Thomas Südhof, a neuroscientist who won a Nobel Prize for his work on synaptic transmission in 2013, and Stanford Professor Alvin Roth who won the 2012 Nobel Prize for his work on the kidney exchange program — are part of its scientific advisory board.