In this post, we’ll share our principles and criteria for investing—not to set rigid boundaries – but to offer clarity about how we think. By sharing this publicly, we aim to give founders insight into what we look for, helping us connect with aligned entrepreneurs faster so they can focus on building. At the same time, we hope to inspire new founders with bold ideas to engage with this space.

We’re one perspective in a diverse ecosystem of investors, and that’s by design. Different criteria mean more founders have a chance to secure funding for their ideas, ensuring a richer and more innovative future for brain and mental health technologies.

5 Investment Areas We’re Excited About

We’re focused on technologies that challenge the status quo and offer meaningful advances in brain health:

1. Therapeutics (i.e. drugs, cell, microbiome therapeutics)

2. Precision Neuroscience Platforms (i.e. scalable tech-based platforms; massive clinical data collection platforms)

3. Therapeutic Medical Devices: (i.e. neuromodulation)

4. Consumer Health (i.e. science-first wearable devices)

5. AI-Enabled Care Platforms (i.e. clinical brain & mental health, preventive health like fitness, mindfulness, nutrition)

1️⃣ Therapeutics (Includes Biotech and Techbio)

What we’re excited about:

• Frontier Technologies: Biotech, TechBio, AI-based drug development—all anchored in direct human neuroscience first data.

• Human-First Neuroscience: Initiate target discovery from patient tissue, brain imaging, or real-world evidence, not mouse models or cells (e.g. Recursion’s NeuroMap based on cells)

• Select Novel Modalities: Psychedelics, cell therapies, microbiome-based treatments, and other frontier approaches that address the brain’s complexity.

• Therapeutics Should Address Neural Complexity: If targeting behavior, very unlikely that a single pathway will be successful

• Risk Tolerance: We embrace new targets and early-stage technology, acknowledging higher risk for a higher reward, while requiring patient-centric evidence.

The Human-First Lens for Neuroscience

The human brain and behavior remain some of the most complex and elusive frontiers of medicine. Traditional pharmaceutical and biotech models have struggled to develop drugs for psychiatric and neurological conditions because they rely heavily on reductionist or purely animal-based data. By contrast, we believe in human-first approaches—those that prioritize patient data and real-world human evidence from the outset, rather than as a downstream validation step.

• Key Principle: Start in patients, not in mice.

  • Example: Verge Genomics has built an AI-driven platform that begins with human brain tissue and genetic data[^1]. By focusing on direct patient data, they aim to increase the odds of clinical success in neurodegenerative diseases.

• Why It Matters:

  • Traditional Approach: Animal or cell models can misrepresent human disease because the complexity of the brain doesn’t translate easily from other species.

  • Human-Centric Approach: Harnesses large patient datasets (e.g., brain tissue banks, genomics, clinical registries), leading to higher translational relevance.

[^1]: Verge Genomics’ core method involves analyzing gene expression in tissue samples from patients with ALS, Parkinson’s, and other neurodegenerative conditions. These findings guide drug target identification before animal studies, rather than after, as in traditional pharma.

Our Focus in Neuro: Human-Centric Frontier Tech

1. Human-Derived Datasets & Tools

   1. Example: Platforms built around clinical data collection, integrated real-time patient monitoring, and large-scale genomic, proteomic, and microbiome datasets.

   2. Why: These resources can drive target discovery and validation directly in humans, sidestepping misleading animal data.

2. Psychedelics with Multi-Target Mechanisms

   1. Why It Fits: Psychedelic compounds (e.g., psilocybin, LSD) operate on multiple receptors and neural circuits simultaneously. This multi-target profile reflects the real-world complexity of consciousness and behavior.

   2. Investment Case: These substances have centuries of real-world “data” (traditional use, plus modern clinical trials). In a sense, they’re already clinically “de-risked,” though regulatory and manufacturing pathways remain challenging.

3. Cell Therapies & Novel Modalities

   1. Rationale: Cells (or engineered cells) carry numerous DNA, RNA, and proteins, potentially addressing the polygenic nature of brain disorders.

   2. Caution: Manufacturing and distribution hurdles are still high. We will monitor how larger pharma players navigate these issues to assess viability at scale[^3].

[^3]: Examples include CAR-T manufacturing processes in oncology; adapting them for brain diseases requires solving blood-brain barrier delivery, immunogenicity, and large-scale production.

Risk and Reward in Novel Targets

As Pre-Seed investors in frontier neuroscience, we accept that many of these therapeutics hinge on first-in-human or novel targets. Our ability to generate alpha lies in identifying these early, transformational ideas before they’re de-risked by conventional research.

• Novel Target Risk

  • Upside: The payoff can be significant if the target truly addresses an unmet need in neuro.

  • Downside: Failure is more likely when relying on unproven targets or incomplete data.

• New Modalities (Microbiome, Cell, Gene Therapies)

  • Upside: Potential for disease-modifying or curative interventions.

  • Downside: Manufacturing complexities, regulatory uncertainties, and high capital requirements.

Red Flags: “Mouse-Driven” Validation

Finally, one of the strongest signals that a frontier drug or TechBio approach is actually not aligned with our human-first lens is when companies use humans merely to validate targets discovered in animals or cell lines.

1. Paper-Driven Targets

   1. Academic Publications: Many high-profile journal articles focus on novel mechanisms in mice. While valuable for mechanistic insight, these often fail in human trials.

   2. Example: TLR4 research[^5] might show promising rodent data but lacks robust human data, making translational success uncertain.

2. Late-Stage “Patient Validation”

   1. What We Avoid: A company that uses reams of animal or simulation data, then brings human cohorts in only to “check a box.”

   2. Why: We believe real-world patient data should be central from the start, shaping every stage of discovery.

[^5]: TLR4 (Toll-like receptor 4) has been implicated in inflammatory processes in rodent models; however, direct evidence in human brain pathology remains sparse or inconclusive.

Conclusion

We invest in frontier, human-centric therapeutics that aim to significantly improve brain and mental health outcomes. We avoid purely animal-based or reductionist approaches that treat human validation as a late-stage afterthought. Our core belief is that by starting with patient data and the real complexity of the human brain, we can build more effective, scalable, and life-changing therapeutics for neurological and psychiatric conditions.

2️⃣ Precision Neuroscience Platforms

What we’re excited about:

• Objective Characterization: Neurological diseases need quantifiable metrics; emerging tech can finally offer them at scale.

• Defensible IP & Data: Proprietary hardware, unique databases, and AI solutions that grow over time.

• Platform, Not One-Off: Solutions must address multiple conditions or endpoints to justify investment.

• Scalable Approaches: Decentralized, user-friendly solutions (wearables, at-home multi-omics) trump facility-only methods.

• Viable Business Model: Reimbursement and clear regulatory pathways are critical to commercial success.

Neuroscience has long struggled with the ability to objectively characterize disease. This lack of objective measures has hindered clinical trials, complicated diagnoses, and undermined the design of effective therapeutic strategies. Historically, diagnostic breakthroughs in neurology have been relatively rare, but when they do occur, they can be transformative (and highly valuable).

Why Now?

Emerging technologies—ranging from advanced neuroimaging and non-invasive devices to wearable sensors and multi-omics—have the potential to radically change how we identify and monitor neurological conditions. Despite ongoing challenges (e.g., Alto Neuroscience[^1] illustrating the complexity of precision neuroscience), we believe the stage is set for a new wave of massive diagnostic platforms that could reshape the field.

[^1]: Alto Neuroscience, a PsyMed portfolio company, focuses on biomarker-based approaches to psychiatry. While promising, they face hurdles in gathering robust, large-scale patient data and translating it into clinically actionable metrics—a key challenge for precision medicine in mental health. They are leading the field in precision psychiatry.

Key Investment Principles for Precision Neuroscience Platforms

1. Defensible Breakthrough Technologies

   1. Definition: The solution should offer a clear competitive advantage—be it proprietary hardware, unique multi-modal data, or a scalable software platform. This should be through a hard technical breakthrough (e.g. new type of power conversion, novel optics technology, etc.)

   2. Why It Matters: This ensures potential monopolistic or first-mover benefits in a competitive, science-driven market.

   3. Example: A wearable EEG device that captures high-fidelity brain signals[^2], combined with a massive, proprietary dataset to train AI algorithms on brain activity patterns– moving forward this will likely not be EEG, but a more quantitative and explainable imaging modality

2. Clinical Data Over “Just” Molecular Data

   1. Premise: AI-driven drug discovery and diagnostics rely on robust, large-scale human clinical data (e.g., real-world data, EHR data, longitudinal patient cohorts, biomarkers, imaging).

   2. Why: Molecular data alone (like genomics or proteomics) is less likely to yield breakthroughs without the real-world clinical context that shapes disease progression.

   3. Example: A platform that integrates patient-reported outcomes, imaging data, and genomic profiles, enabling more accurate biomarker discovery than pure lab-based molecular screens.

3. Platform Diagnostics, Not One-Off Tests

   1. Scope: A single blood test for Parkinson’s disease is not a compelling investment if it lacks potential for broader application.

   2. Why: Platform-based solutions can address multiple neurological or psychiatric conditions, increasing ROI and impact.

   3. Example: A biomarker detection platform that can pivot to test for various neurodegenerative diseases (e.g., Alzheimer’s, ALS, Parkinson’s) by simply adjusting reagents or analytical parameters[^3].

4. Scalability Is Essential

   1. Neuroimaging Constraints: If a diagnostic relies on specialized equipment and expert operators (e.g., advanced MRI protocols), it limits scalability.

   2. Exception: Pairing an imaging-based diagnostic with a specific therapeutic program (companion diagnostics) can work if the solution is co-developed for clinical trials or precision treatments.

   3. Caveat: Neuroimaging is best leveraged to show target engagement or surrogate endpoints, rather than forming the entire basis of FDA approval.

5. Wearable & At-Home Models: Non-invasive, user-friendly devices (e.g., wrist-worn sensors, mobile EEG, or saliva-based multi-omics) allow for decentralized data collection and larger patient pools.

   1. Why It Matters: Broad adoption drives faster data generation, deeper insights, and more robust predictive models.

6. Clear Business Model & Reimbursement Path

   1. Rationale: Even a clinically game-changing test might sturggle if payors or health systems won’t reimburse it.

   2. Example: Blood tests for early Alzheimer’s detection show promise scientifically but face hurdles in pricing, coding, and insurance coverage[^4].

   3. Investor Perspective: Look for diagnostic tools with well-defined CPT codes, proven cost-effectiveness, or partnerships with payors/pharma to ensure sustainable revenue.

[^2]: Wearable EEG devices are advancing in signal fidelity and artifact rejection, allowing real-time brain monitoring outside of clinical settings. Companies like Cognito Therapeutics (for neuromodulation/EEG) and Kernel (for functional near-infrared spectroscopy) showcase innovations in this space.

[^3]: Some platforms leverage multiplex immunoassays or next-generation sequencing that can switch “targets” by reconfiguring the assay design, providing flexible pipelines for new indications.

[^4]: Example: C2N Diagnostics developed a blood test for Alzheimer’s disease measuring specific amyloid-beta isoforms. Despite strong clinical potential, widespread adoption depends on Medicare and private insurance coverage decisions.

1. Technology + Data = Defensibility

   1. Combine novel hardware or software with a robust data engine. Aim for large, proprietary datasets and the capacity to grow them over time.

2. Broad Platform vs. Narrow Use Case

   1. Favor multi-indication or multi-parameter approaches. Single tests or one-disease solutions face limited growth trajectories and higher commercial risk.

3. Scalability

   1. Seek solutions that decentralize data collection, enabling broader patient participation and faster iteration. Specialized, facility-based scans or tests often hit adoption bottlenecks.

4. Reimbursement & Regulations

   1. Evaluate the path to regulatory approval, alignment with existing reimbursement structures, and the viability of forging new payment codes if needed.

5. Integration with Therapeutics

   1. Companion diagnostics can be strategically valuable—especially in neuro. If imaging or biomarker-based solutions pair with drugs, they can demonstrate target engagement or efficacy in trials, accelerating the development timeline.

Conclusion

We believe the convergence of non-invasive monitoring, wearable tech, multi-omics, and AI is poised to crack open the diagnostics dilemma in neuroscience. The next generation of game-changers in this space will be platform-based, human-centric, and highly scalable, offering robust data to fuel drug discovery, patient stratification, and clinical decision-making. By focusing on defensible technologies, large-scale clinical data, and clear reimbursement pathways, we can capture immense value and drive meaningful improvements in how neurological diseases are diagnosed and managed. Note: There will likely be room for one major winner per technology type.

3️⃣ Neurotech & Brain Computer Interfaces

• Non-Invasive Advantage: Potentially fewer side effects, more targeted therapies, and adaptive dosing in real time.

• Complexity as a Feature: Brain responses to energetic or mechanical modulation inherently engage multiple pathways, allowing for multifaceted therapeutic effects.

• Scalable, Patient-Centric: Solutions must minimize specialized operational requirements, prioritize user experience, and aim for at-home or outpatient usage.

• Technical + Clinical Validation: Breakthrough engineering plus rigorous trials yields defensible IP and real-world efficacy data.

• Regulatory & Reimbursement: Ensure a strategic path to FDA approval and a viable reimbursement model to achieve wide adoption.

As non-invasive technologies continue to evolve, their potential to improve brain health is increasingly evident. Compared to pharmacological strategies, well-designed devices can offer more targeted interventions, fewer systemic side effects, and the ability to dynamically adjust treatment to a patient’s changing condition. From transcranial focused ultrasound (tFUS) to transcranial magnetic stimulation (TMS) and beyond, these interventions harness the brain’s complex response to energetic or mechanical stimuli—rather than a single molecular target.

Core Principle

We value technical breakthroughs with clinical validation, led by charismatic engineers who are deeply motivated by patient outcomes. This combination accelerates real-world application and ensures a patient-first lens is upheld from design to deployment.

Major Opportunities for Devices

1. Novel Quantification of Brain Health

   1. Rationale: Advancements in sensors, imaging, and signal processing have opened up new ways to measure brain activity and physiology non-invasively.

   2. Example: Next-generation EEG or MEG systems[^1] that can parse subtle biomarkers of cognitive decline or early Alzheimer’s disease in real time.

   3. Investment Angle: Platforms that identify disease states or track therapeutic response faster, cheaper, and more accurately than current clinical gold standards.

2. Treating Conditions with Few Effective Drugs

   1. Premise: Some neurological and psychiatric conditions have limited or lackluster pharmacological options.

   2. Example: TMS for treatment-resistant depression[^2], offering durable results for patients who have tried multiple medications without relief.

   3. Future Outlook: Devices can fill therapeutic gaps or provide add-on therapies, sometimes displacing or reducing reliance on medications with heavy side effects.

3. Neuromodulation for Inflammatory Conditions

   1. Opportunity: Non-invasive devices that stimulate peripheral nerves or targeted brain regions might modulate immune responses, circumventing the need for immunosuppressants or steroids.

   2. Example: Vagus nerve stimulators[^3] that have shown promise in reducing inflammation for autoimmune disorders, with fewer systemic adverse effects compared to biologic drugs.

   3. Why It Matters: Many inflammatory conditions also have neurological or psychosomatic components, making a neuromodulation approach doubly compelling.

4. At-Home Wearable Treatment Devices

   1. Scalability: Portable or wearable neurostimulators (headbands, wristbands, or patches) allow patients to receive therapy on their own schedule.

   2. Example: A wearable tFUS headband[^4] that delivers low-intensity focused ultrasound to targeted brain regions.

   3. Aesthetics are essential to drive adoption, be inclusive to patients, and have long lasting market value

   4. Investor Note: Home-based, user-friendly devices can generate real-world patient data at scale and open the door to new DTC (direct-to-consumer) or reimbursable clinical models.

5. Closed-Loop Neuromodulation

   1. Concept: Devices that sense neural activity and adjust stimulation parameters in real time, akin to a thermostat for brain function.

   2. Example: Deep brain stimulation (DBS) systems with real-time EEG feedback[^5], customizing pulses to minimize side effects and enhance efficacy (e.g., for Parkinson’s or epilepsy).

   3. Advantage: Adaptive stimulation can deliver “just enough” intervention, lowering battery usage and side effects while improving outcomes.

6. Individualized Dosing Strategies

   1. Why: Patients vary widely in their neurophysiological responses to device-based therapies. Tailoring intensity or frequency of stimulation to personal thresholds can optimize outcomes.

   2. Example: Systems that measure maximum evoked potential in motor cortex and then apply a percentage of stimulation for each patient’s session[^6].

   3. Impact: Personalized treatment protocols may lead to fewer adverse effects and faster clinical improvement.

Key Investment Criteria

1. Technical Breakthrough + Clinical Rigor

   1. Definition: Look for devices that demonstrate a clear, protective technological edge (e.g., unique IP around ultrasound targeting, advanced signal filtering, or custom hardware design) combined with robust clinical validation (pilot trials, published data, KOL endorsements).

   2. Why: Differentiation in the med-tech space typically comes from combining engineering innovation with reproducible outcomes in patients.

2. Scalable and Accessible

   1. Premise: Devices should be convenient, with minimal specialized training or facility requirements.

   2. Example: Portable TMS solutions (e.g. PsyMed Portfolio company, Motif Neurotech) that can be deployed in primary care or outpatient clinics, rather than costly hospital-based installations.

   3. Business Model: Treatments that are easy to administer and maintain can expand access to underserved areas and potentially integrate with telemedicine or remote monitoring solutions.

3. Patient-Centric Design

   1. Rationale: Comfort, safety, and intuitive use are paramount to ensure adherence and foster positive patient experiences.

   2. Example: Ergonomic headsets with app-based guidance for home therapy sessions, offering user-friendly interfaces and clear tracking of progress.

4. Charismatic, Mission-Driven Founders/Engineers

   1. Why: Hardware-intensive startups often face longer development cycles and regulatory complexity. Passionate, resourceful teams are more likely to navigate these hurdles successfully and pivot when needed.

5. Regulatory Path and Reimbursement

   1. Considerations: FDA device classifications, CE marks, CPT codes for reimbursement, and potential for coverage by insurers or major healthcare providers.

   2. Impact: A robust regulatory strategy can differentiate a device startup from competitors stuck in extended clinical trial phases or lacking a clear go-to-market plan.

Conclusion

Medical devices for brain health represent a high-impact category in frontier neuroscience. By targeting functional (rather than purely molecular) aspects of disease, these technologies embrace the complexity of the human brain while improving patient quality of life—often with fewer side effects than pharmaceuticals. We see significant opportunities in non-invasive and minimally invasive devices, especially those with closed-loop capabilities, scalable form factors, and a clear path to clinical integration.

[^1]: Advanced EEG/MEG technologies can offer real-time functional mapping of brain networks, potentially detecting subclinical biomarkers of various neurological disorders.

[^2]: TMS for depression has gained FDA clearance; companies like NeuroStar and BrainsWay have significantly expanded the market for device-based mental health treatments.

[^3]: Vagus nerve stimulators (e.g., from SetPoint Medical) are being explored for rheumatoid arthritis, Crohn’s disease, and other inflammatory disorders, showing early promise in reducing chronic inflammation.

[^4]: tFUS headbands or helmets are under investigation for conditions like essential tremor and major depression; low-intensity pulses can be focused on specific cortical regions.

[^5]: Closed-loop DBS systems (e.g., Medtronic Percept) use embedded sensing to tailor stimulation, reducing side effects like dyskinesias in Parkinson’s disease.

[^6]: “Dosage” of neuromodulation can be calibrated using motor evoked potentials (MEPs), then adjusted for individual neuroplastic responses over time.

4️⃣ Consumer Health

• Complexity Is an Asset: The best interventions embrace multi-pathway biological processes (akin to exercise, diet, sleep).

• Differentiation via Science: Strong, defensible IP and rigorous validation are crucial in a market often undermined by snake oil.

• Data Depth & Utility: High-resolution data in real-world contexts fuels unique insights and potential pivot to clinical realms.

• Supplements Require Rigor: Avoid commoditized offerings; demand robust science to stand out.

• Platform & Path to Clinical: Consumer products that can evolve into regulated solutions (or vice versa) offer significant upside.

• Payor Strategy: Identify who foots the bill—direct consumers or third-party payors—early, as it shapes development and commercialization.

In many ways, the best interventions for health—exercise, diet, and sleep—succeed precisely because they engage numerous biological pathways simultaneously. Rather than narrowing their scope to a single pathway, these interventions embrace the inherent complexity of the human body. New consumer-facing health products and services can follow a similar playbook, provided they remain science-first and build around strong technical innovation—whether that’s in hardware, data science, or novel molecular insights.

However, the supplement market shows how credibility can erode without rigorous scientific underpinnings. The industry is largely unregulated, making it easy for low-efficacy or non-replicable products to flood the market. To stand out, consumer products should offer both objective measurements (e.g., biomarkers or robust clinical data) and tangible health benefits, ideally validated by academic experts or clinical trials.

Key Principles for Consumer Health Investments

1. Embrace Complexity, Don’t Reduce It

   1. Rationale: Multi-pathway interventions are often the most powerful (e.g., exercise affects metabolism, neurology, cardiovascular function).

   2. Investment Angle: Solutions that leverage this holistic approach—combining data from wearables, AI-driven insights, or integrative health platforms—are likely to be more sustainable than single-pathway “silver bullets.” Ideally these are quantifiable, and thus comparable to clinical grade literature, responsive to interventions, and reliable within AND between subjects.

   3. Example: A wearable that tracks sleep architecture, stress responses, and physical activity in one unified dashboard[^1].

2. Science-First, Technically Defensible

   1. Premise: Consumer health products with credible, peer-reviewed science or novel technology (device innovation, new molecular discovery, advanced AI) are more likely to differentiate themselves.

   2. Why It Matters: Building a defensible moat in consumer health often requires high-bar validation—ensuring that imitators can’t simply copy the idea overnight.

   3. Example: A microbiome analysis startup using proprietary gene-sequencing and AI[^2] to recommend dietary changes or personalized probiotics that have undergone initial clinical validation.

3. High-Resolution Data in Unique Contexts

   1. Opportunity: Consumer tech that collects granular, real-world health data creates robust datasets, leading to proprietary insights.

   2. Example: A smart ring capturing continuous physiological metrics (e.g., Oura) or a next-generation patch measuring stress hormones on the skin (e.g. Kolibri)[^3].

   3. Investor Perspective: Proprietary datasets can fuel iterative improvements, predictive analytics, and potential transitions into regulated, clinical applications.

4. Supplements: Extremely High Threshold

   1. Context: The unregulated supplement space can undermine credibility. To succeed, products must demonstrate rigorous scientific backing, ideally with clinical data.

   2. Example: Cambiotics; identification of a specific microbial strain that isolates environmental toxins. Confirmed with mass spectrometry and bacterial genome sequencing, these strains are novel and scalable.

   3. Why: This level of rigor helps distinguish effective solutions from the noise and builds long-term consumer trust.

5. Platform Capacity & Path to Clinical Trials

   1. Reasoning: Great consumer products can evolve into clinically validated solutions—be it surrogate endpoints, companion diagnostics, or full-fledged medical devices. ARPA-H is now investing heavily in decentralized clinical trials and the FDA is developing infrastructure for Real World Evidence (RWE) to help with determining therapeutic safety and efficacy in the real world.

   2. Example: A wearable for general wellness that later secures an FDA indication for monitoring seizure activity in epilepsy patients[^4].

   3. Value: This optionality can significantly increase a company’s upside, offering both consumer revenue and the potential for insurance reimbursement or partnership with pharmaceutical companies.

6. Consumer → Regulatory or Regulatory → Consumer

   1. Decision Point: Founding teams must decide whether to start with consumer adoption and then seek regulatory clearance or pursue a formal approval path first to establish credibility.

   2. Team Fit: Technical founders with deep R&D expertise might choose a regulatory-first path, whereas teams with strong marketing DNA may launch consumer products and use academic partnerships to bolster credibility.

   3. Caution: A pure consumer launch without meaningful validation can lead to skepticism; regulatory approval without a viable consumer strategy can slow go-to-market traction.

7. Microbial & Environmental Approaches

   1. Complexity: Altering microbial communities (e.g., gut, skin) or mitigating environmental toxins touches many biological systems at once.

   2. Example: A probiotic regimen designed to shift a user’s microbiome, validated by breath or stool test analytics[^5]. (e.g. H. Pylori isotope breath test is a clinical assay)

   3. Investor Lens: Understand the multifaceted nature of these interventions and ensure there is a well-defined business model and data to show real clinical or consumer benefit.

8. Who Will Pay for It?

   1. Business Model: If an offering is cash-pay, it competes with other out-of-pocket health purchases—such as fitness devices, nootropics, or wellness apps.

   2. Reimbursement: If a product aims for insurance coverage, expect longer sales cycles and the need for proven cost-effectiveness.

   3. Strategic Insight: Early clarity on reimbursement vs. consumer pay sets the tone for go-to-market, product design, and required clinical evidence.

Conclusion

Consumer health is a vibrant arena for innovation, but also crowded with unverified products. Your investment strategy emphasizes science-driven, high-complexity approaches that can truly move the needle on brain and overall health. By focusing on technical defensibility, clinical validation, and a flexible path to regulatory or consumer markets, these ventures can capture both commercial success and genuine patient/consumer impact.

[^1]: The WHOOP band, Apple Watch, and Oura Ring all exemplify consumer wearables that integrate multiple metrics (e.g., sleep stages, HRV) into daily feedback loops.

[^2]: Companies like Viome or DayTwo analyze gut microbiome data to create personalized nutrition plans, though the depth and quality of clinical validation vary.

[^3]: Emerging wearable patches are exploring cortisol level monitoring as a real-time stress biomarker for consumer health.

[^4]: Empatica’s Embrace watch originally targeted seizure monitoring, but also found broader consumer traction as a stress-tracking wearable.

[^5]: Some probiotic or postbiotic companies use at-home test kits to measure changes in microbiome composition over time, though clinical efficacy remains a moving target.

5️⃣ AI Enabled Care Platforms

• Complex Conditions Need Human Touch: AI augments but rarely fully automates.

• Biomarker-Driven Models: Rich, unique data (voice, wearables, daily logs) enable better predictive algorithms.

• Pharma and Payor Partnerships: Providing clinically relevant insights or patient stratification can unlock significant revenue.

• At-Home Care: AI addresses shortages in traditional caregiving, especially for mental health or senior populations.

• Proprietary Data or Technical Breakthrough: Must have a clear moat in an increasingly crowded AI-health field.

• Behavior Change Focus: Tools that keep patients engaged and adherent are more likely to deliver real outcomes.

Healthcare challenges often require nuanced, behavior-focused interventions—especially in neurological and psychiatric domains. AI can help scale human-driven care by facilitating behavior change, automating low-complexity tasks, and personalizing interventions. However, total automation may be premature in areas lacking objective diagnostics or well-established treatment algorithms.

Core Concept

The most promising AI solutions balance machine-driven insights with a clinician (or caregiver) in the loop, leveraging unique data sets (e.g., wearables, voice data, daily logs) to improve outcomes without replacing the human touch altogether.

1. Person-in-the-Loop, Not Machine-Only

   1. Context: Many psychiatric or neurological conditions defy rigid, algorithmic protocols.

   2. Example: Jimini’s, a PsyMed portfolio company, approach to AI-driven behavior change[^1], which offers nudges and reminders based on patient-reported mood, but still encourages clinician or coach oversight for complex decisions.

   3. Rationale: Automation excels when objective data is abundant and validated care pathways exist. In ambiguous clinical areas (e.g., mild depression, early dementia), human expertise remains vital.

2. High-Resolution, Clinically Relevant Data

   1. What It Is: Data from wearables, voice analysis, or real-world monitoring that can be fed into predictive models.

   2. Why It Matters: These biomarkers can inform clinical trial design, identify early warning signs, and refine patient stratification for pharma or care providers.

   3. Example: An app that collects speech patterns, sleep metrics, and heart rate variability to predict relapse in bipolar disorder[^2].

3. Path to Pharma and Payor Partnerships

   1. Premise: Large, longitudinal datasets are often invaluable to pharma companies seeking validated biomarkers and improved patient segmentation.

   2. Opportunity: A venture can build a robust business model by licensing data insights, co-developing tools for clinical trials, or offering real-time patient monitoring solutions.

   3. Example: Yale researchers recently combined wearable data with genomic data in order to increase diagnostic accuracy and more closely link genetics with behavioral features. This might enable

4. Novel Technical Breakthrough or Proprietary Data

   1. Requirement: For an AI platform to stand out, it should either harness a truly unique dataset or employ a technical breakthrough in model architecture, personalization, or integration with existing systems.

   2. Why: Healthcare is data-rich, but large incumbents (e.g., EHR vendors, major hospital chains) have significant clout. A startup needs a major differentiator to succeed.

   3. Example: A quantitative EEG dataset pre and post treatment combined with real-time AI analysis that no other institution can replicate, enabling more precise diagnosis or therapy optimization.

5. Automated Diagnosis vs. Augmented Diagnosis

   1. Where We’re Skeptical: Completely AI-driven diagnostic tools for psychiatric or neurological conditions that rely on subjective criteria (e.g., DSM-based categories) and lack objective biomarkers. While diagnostic accuracy for primary care, internal medicine, and emergency medicine is higher for AI than clinicians (See JAMA Network Open study), neuro related diagnostics are likely a ways away. This is due to the inability for AI systems to adequately capture diagnosis relevant data-- if you solve this specific problem, talk to us!

   2. Preferred Approach: Augmented tools that enhance clinician decision-making (e.g., highlight patterns, suggest treatment adjustments), leaving critical judgment to a trained professional.

6. Behavior Change as a Key Factor

   1. Context: Many chronic conditions hinge on patient lifestyle and adherence to therapy plans. AI can be a powerful motivator if deployed correctly (e.g., personalized coaching, real-time feedback, gamification).

   2. Example: An AI-driven CBT (Cognitive Behavioral Therapy) app that integrates daily routines, wearable data, and mood tracking to tailor therapeutic exercises, but still offers on-demand clinical check-ins.

Investment Considerations

1. Validation Path

   1. Clinical Trials: Tools intended for regulated indications should demonstrate efficacy through clinical trials or robust observational data.

   2. Real-World Evidence: For purely consumer-facing tools, large user bases and peer-reviewed outcome studies can replace traditional trials but must still show rigorous, reproducible benefits.

2. Reimbursement & Business Model

   1. Potential Models:

      1. Subscription (employers, health systems).

      2. Partnerships with Pharma (licensing data, companion diagnostics).

      3. Direct-to-Consumer (less typical for advanced AI platforms, but possible if trust is established).

   2. Example: A care coordination platform that charges health systems per-member-per-month for improved patient engagement and reduced readmissions (for example ways of enhancing intensive outpatient programs (IOPs))

3. Integration with Existing Workflows

   1. Reasoning: AI can complement human clinicians, but must integrate smoothly with established EHR systems and clinical processes.

   2. Challenge: Healthcare IT infrastructure is notoriously fragmented, so solutions that simplify providers’ experiences tend to gain traction faster.

4. Team & Technical Edge

   1. Emphasis: Founders need deep AI expertise and knowledge of healthcare delivery or neuroscience. Pure tech teams without clinical understanding may overpromise or fail to gain provider trust.

   2. Example: An AI startup launched by a neurologist-ML engineer duo who already understand clinical data collection constraints and physician workflow.

Conclusion

AI-enabled care platforms hold tremendous potential for scaling complex interventions, particularly in mental health and neurology, where behavior change and ongoing monitoring are paramount. Success hinges on high-resolution, proprietary data, a clinician-in-the-loop design, and a validated approach that either fits into existing reimbursement frameworks or secures strong pharma partnerships.

[^1]: Jimini is an early-stage AI startup that guides patients between doctor visits, employing 24/7 therapy access for patients that still allows a physician to oversee complex cases.

[^2]: Some companies have explored voice biomarkers (e.g., Kintsugi) to detect emotion or stress markers correlated with depression or bipolar disorder relapse.

[^3]: AI chatbots for cognitive decline are an emerging field, with startups exploring ways to remind patients of medication schedules, next appointments, or daily tasks while monitoring changes in language patterns.

PsyMed Ventures is evolving to back the next wave of transformative technologies in brain health. By focusing on human-first neuroscience and embracing high-risk, high-reward opportunities, we aim to support innovations that can truly change the way we understand and treat the brain. Sharing our investment principles helps us connect with visionary founders who share our mission and accelerates the journey to a healthier, more innovative future in neuroscience. We’re excited to see where this journey takes us—and the impact it will have.

Share PsyMed Ventures Newsletter