New Lumos AI platform targets precision in mental health drugs

Headlamp Health has announced the launch of Lumos AI®, a comprehensive, intelligent platform designed to help drug developers navigate the biological and behavioural complexity of neuroscience drug development.

Unlike artificial intelligence tools focused primarily on workflow automation or trial operations, Lumos AI operates as a decision-support layer. It applies clinical logic and pattern recognition to biological, behavioural and clinical signals, enabling development teams to make better-informed decisions earlier in the drug development process.

“Mental health drug development has long operated on outdated assumptions, treating complex conditions with one-size-fits-all approaches,” said Zak Williams, advisor for Headlamp Health. “We’re now at an inflection point where advances in technology make precision possible in a way that wasn’t even imaginable five years ago. That shift is critical to delivering the right care to the right people at the right time.”

Why neuroscience lags behind oncology

In recent years, oncology has led the shift towards precision medicine, grounding drug development in well-characterised biological markers and clear responder–non-responder frameworks. Neuroscience, by contrast, has been slower to adopt similar approaches, largely due to the complexity of brain-related disorders.

Traditional neuroscience trials have struggled to account for patient heterogeneity, subjective symptom reporting and powerful placebo effects. As a result, complex biological signals are often reduced to population averages that obscure meaningful individual differences. This has contributed to high failure rates, as trials are unable to reliably identify the right patient populations or detect early signals of efficacy.

Lumos AI is designed to address these issues by providing scientists with a more precise understanding of treatment response. The platform helps teams identify appropriate patient populations and detect meaningful signals sooner, reducing uncertainty around trial design and responder identification.

“We built Lumos AI to address two fundamental questions: which patients are most likely to benefit from a given therapy, and which new or existing therapies are most likely to work for a given patient subtype,” said Erwin Estigarribia, CEO of Headlamp Health. “Lumos AI helps pharmaceutical development teams ask better questions earlier by understanding variability rather than relying on volume alone.”

Key capabilities across development

Lumos AI supports decision-making across the neuroscience drug development lifecycle. Its core capabilities include identifying responder and non-responder patient subtypes, refining trial strategy through earlier insight into inclusion criteria and study design and modelling how patients change over time. The platform also provides portfolio-level decision support, helping companies de-risk early- and mid-stage development programmes.

Moving beyond static trial design

Recent advances in artificial intelligence, coupled with access to large-scale longitudinal real-world data, have created new opportunities to move beyond static trial designs in neuroscience. Lumos AI offers a longitudinal view of patient response that better reflects how conditions evolve over time and how treatments perform in real-world scenarios.

“Psychiatry has settled for a ‘responder’ definition that effectively means a patient is only 50 percent less miserable. We wouldn’t accept a 50 percent reduction in tumour load as a success in oncology, and we shouldn’t accept it here,” said Dr Charles B. Nemeroff, chair of psychiatry at UT Austin and advisor to Headlamp Health. “Headlamp’s approach focuses on remission, getting patients actually well, not just slightly better, by using continuous data to guide them to the right treatment faster.”

The launch of Lumos AI coincides with the appointment of Williams and Nemeroff to Headlamp Health’s Board of Advisors, strengthening the company’s academic and advocacy expertise as it progresses precision neuroscience.