Gidi Littwin, a primary co-inventor of Apple’s transformative FaceID and Vision Pro tracking technologies, has transitioned from the world of consumer electronics to the frontier of medical neuroscience. His new startup, Hemispheric, recently announced a successful $52 million funding round aimed at scaling a proprietary artificial intelligence model designed to decode the complex electrical patterns of the human brain. By leveraging the same deep learning principles that allow iPhones to recognize faces and headsets to track hand movements, Hemispheric seeks to provide a non-invasive, objective window into cognitive health, potentially transforming the way disorders such as PTSD, Alzheimer’s, and depression are diagnosed and managed.
The funding round, which included participation from prominent American and Israeli venture capital firms as well as high-profile individual investors like early Uber-backer Howard Morgan, marks a significant milestone for the Tel Aviv and Boston-based company. The capital infusion is earmarked for the expansion of Hemispheric’s data collection operations, the pursuit of rigorous regulatory approvals, and the development of specialized hardware tailored for machine learning applications.
The Genesis of Hemispheric: From Silicon Valley to Synaptic Mapping
The origins of Hemispheric date back to 2020, a year of transition for Gidi Littwin. Having spent years at Apple contributing to some of the most sophisticated computer vision and biometric systems in the world, Littwin was seeking a new challenge that applied high-scale data modeling to human health. His path crossed with Hagai Lalazar, a neuroscientist who had been quietly developing AI frameworks to study brain activity without the necessity of surgical implants or invasive procedures.
Lalazar had already recognized a fundamental bottleneck in modern neurology: while the brain generates massive amounts of electrical data, the tools to interpret that data in a clinically actionable way were lagging behind. Before connecting with Littwin, Lalazar had interviewed approximately 75 potential co-founders, searching for a partner who possessed both the technical acumen to handle "frontier" AI models and the commercial experience to navigate the complexities of global product launches.
The synergy between the two founders was immediate. Littwin’s experience at Apple—specifically in managing the "massive data collection operations" required to train FaceID and the Vision Pro’s hand-tracking algorithms—provided the blueprint for Hemispheric’s strategy. Just as FaceID required hundreds of thousands of diverse subjects to ensure reliability across different lighting conditions and facial structures, decoding the brain would require a dataset of unprecedented scale and diversity.
Bridging Consumer Tech and Medical Neuroscience
During his tenure at Apple, Littwin learned that the efficacy of a deep learning model is directly proportional to the quality and volume of the training data. For the Vision Pro, this meant capturing how thousands of different people move their hands in three-dimensional space. At Hemispheric, the "subject" is the electrical activity within the skull.
Traditional neurology has often relied on subjective measures. For decades, the diagnosis of mental health conditions and neurodegenerative diseases has been heavily dependent on patient questionnaires, behavioral observations, and the clinical intuition of doctors. While valuable, these methods lack the objective precision of a blood test or an MRI. Hemispheric’s goal is to bridge this gap by treating brain activity as a data-rich signal that can be decoded using statistical analysis.
The company’s "frontier model" functions similarly to large language models (LLMs) like GPT-4. Where an LLM predicts the next word in a sentence based on patterns in vast amounts of text, Hemispheric’s model infers brain function and health status by analyzing electrical signals (EEG). To achieve this, the company has already amassed what they call their "most prized possession": a dataset comprising 250,000 hours of brain activity from 100,000 paid volunteers across diverse geographical hubs, including Tel Aviv, Boston, and various locations in Asia.
The Data Collection Engine: Games for the Brain
To ensure the AI model understands how a healthy brain functions versus one affected by pathology, Hemispheric’s subjects engage in a series of digital activities. These tasks, which resemble simple video games, are meticulously designed to activate specific neural circuits associated with memory, attention, emotional regulation, and motor control.
While a participant interacts with these "games" on a tablet, they wear a lightweight, non-invasive EEG headset. This allows the system to correlate specific cognitive challenges with real-time electrical responses. By aggregating this data across a massive population, Hemispheric has built a "generalized model" of the human brain.
The company has already begun testing this model on specific subsets of the population, including individuals previously diagnosed with schizophrenia, depression, and Post-Traumatic Stress Disorder (PTSD). According to the founders, the model has demonstrated a high degree of accuracy in identifying the "neural signatures" of these conditions, moving the diagnostic process from the realm of subjective description to objective data analysis.
Clinical Ambitions: PTSD and the Alzheimer’s Challenge
Hemispheric’s immediate roadmap is focused on obtaining regulatory clearance. The company plans to submit its first diagnostic product—specifically designed for the study and diagnosis of PTSD—to the U.S. Food and Drug Administration (FDA) in early 2025. If successful, this would pave the way for a public rollout by 2027.
Beyond PTSD, the company is deeply invested in the fight against neurodegenerative diseases. Hemispheric is currently conducting clinical studies to determine if its AI model can not only diagnose but also predict the onset of Alzheimer’s disease. Current diagnostic tools for Alzheimer’s often identify the disease only after significant cognitive decline has occurred. An AI-driven tool capable of detecting subtle shifts in brain electrical activity years before symptoms manifest could revolutionize early intervention and pharmaceutical research.
The vision, as articulated by Hagai Lalazar, is to make brain health monitoring as routine and accessible as a standard blood panel. "The future that we envision is one where this is akin to a blood test," Lalazar stated in a recent interview. He emphasized that the eventual hardware—a refined, low-cost EEG headset—is intended to be distributed widely, reaching beyond specialized hospitals into local mental health clinics and even psychologists’ offices.
Specialized Hardware: Reimagining the EEG
One of the most ambitious aspects of Hemispheric’s strategy is the development of its own proprietary brain scanners. While traditional EEG machines have existed for nearly a century, Littwin argues they were never designed with modern machine learning in mind.
"These devices were never built for machine learning and definitely not deep learning," Littwin noted. Standard EEG equipment often suffers from "noise" or signal interference and can be cumbersome to set up, requiring conductive gels and long preparation times. Hemispheric’s goal is to create a "dry" sensor headset that provides high-fidelity data specifically formatted for AI consumption. This hardware-software integration is a hallmark of the Apple philosophy that Littwin is now applying to the medical field.
Market Context and the AI Healthcare Revolution
Hemispheric enters the market at a time of intense competition and rapid innovation in the "MedTech" space. Major AI players like OpenAI and Anthropic have signaled increasing interest in healthcare applications, while established pharmaceutical giants are seeking digital biomarkers to improve the success rate of clinical trials.
The use of AI in diagnostics is already showing results in other fields. In Europe, AI-assisted tools for detecting lung cancer on CT scans are speeding up triage and treatment. However, the brain remains the "black box" of medicine. While companies like Elon Musk’s Neuralink are focusing on high-bandwidth, invasive brain-computer interfaces (BCIs) that require surgery, Hemispheric is betting on the scalability of non-invasive technology. By avoiding the risks and costs of surgery, Hemispheric aims for a much larger total addressable market, targeting the hundreds of millions of people worldwide affected by mental health and neurological conditions.
Broader Implications and Future Outlook
The $52 million in new funding will allow Hemispheric to expand its workforce in the United States and deepen partnerships with healthcare organizations and pharmaceutical firms. For drug developers, Hemispheric’s technology offers a potential way to measure the efficacy of new psychiatric medications in real-time by observing changes in brain activity patterns.
As the company moves toward its 2025 FDA submission, it faces several challenges. Regulatory bodies like the FDA maintain a high bar for "Software as a Medical Device" (SaMD), requiring extensive proof of safety, efficacy, and the elimination of algorithmic bias. Hemispheric’s diverse dataset of 100,000 individuals is a strategic asset in this regard, intended to ensure the AI performs reliably across different ethnicities and age groups.
If Hemispheric succeeds, it could mark the beginning of a new era in "computational psychiatry." By turning the brain’s electrical "noise" into a clear diagnostic signal, the company seeks to de-stigmatize mental health through objective data and provide clinicians with the tools needed to select the most effective interventions for their patients. The transition of Gidi Littwin from tracking fingers on a screen to tracking the fundamental health of the human mind represents a significant pivot in the application of consumer-grade AI expertise to the most pressing challenges in global health.
