As the pipeline of obesity drugs continues to expand, researchers are trying to better understand why patients respond differently to the same treatments. GLP-1 drugs have revolutionized obesity treatment, but variable response and high discontinuation rates remain major challenges.
One area of increasing interest is the obesity phenotype associated with emotional hunger, or emotional and reward-driven eating behaviors. Despite its clinical relevance, it remains difficult to study this phenotype on a large scale due to the lack of measurable biological markers.
A new study from Phenomix Sciences investigated whether machine learning and genetic risk scoring can identify biological signals associated with emotional hunger. The findings, presented at the 2026 Pacific Biocomputing Symposium, demonstrated the feasibility of combining genetic and behavioral data to assess risk in obese patients.
Dr. Timothy O’Connor, chief technology officer at Phenomics Sciences, spoke about this discovery and its potential impact on obesity drug development and precision medicine. With more than 20 years of experience spanning bioinformatics, data science, and software engineering, O’Connor has previously held positions at Microsoft, Illumina, and CareDx. At Phenomix Sciences, he leads the development of the company’s machine learning platform to analyze biological data related to obesity phenotypes across clinical and data science teams.
The challenge of defining emotional hunger
Emotional hunger is characterized by eating in response to emotional triggers rather than physiological hunger signals. Patients with this phenotype are often affected by stress, mood, and reward-seeking behavior. Previous research suggests that their response to obesity treatment may be different than other patient groups.
However, despite its clinical relevance, emotional hunger remains difficult to quantify biologically.
“The emotional hunger phenotype is more difficult to define because it lies at the intersection of biology, behavior, and environment,” O’Connor says.
The emotional hunger phenotype is more difficult to define because it lies at the intersection of biology, behavior, and environment.
Traditional approaches primarily relied on questionnaires and self-report tools. Although convenient, these methods only capture a snapshot in time and do not necessarily reveal the underlying biological susceptibility. As a result, emotional hunger remains difficult to study at scale or incorporate into drug development and clinical research.
This issue has become even more relevant as obesity treatment options expand and researchers seek to tailor treatments to specific patient groups.
Machine learning and hidden biological signals
Phenomix Sciences uses machine learning to identify subtle biological patterns associated with emotional hunger. The company’s Machine Learning Genetic Risk Score (ML-GRS) framework aggregates signals across multiple genetic pathways, rather than focusing on a single biomarker.
“Machine learning is extremely valuable because it helps us find subtle signals across multiple genetic variations that are difficult to interpret on their own,” O’Connor explained. “The ability to aggregate signals across pathways, particularly those related to mood regulation and reward processing, allows us to identify patterns associated with emotional hunger,” he added.
Machine learning is extremely valuable because it helps find subtle signals across multiple genetic variations that are difficult to interpret on their own.
The company’s ML-GRS approach combines genetic and behavioral information to build a more complete picture of obesity phenotypes. Genetic variants associated with anxiety, depression, and reward-driven eating will be analyzed along with targeted behavioral data.
“ML-GRS models integrate different layers of data,” O’Connor says. “On the genetic side, we are looking at different variations across pathways related to anxiety, depression, and reward-driven eating to create a score that reflects an individual’s biological susceptibility.”
Importantly, the behavioral component is streamlined rather than relying on lengthy questionnaires.
“The next step is to focus on a small number of specific questions rather than a complete survey and amplify that signal with targeted behavioral data,” he explained.
O’Connor said this hybrid model allows researchers to distinguish between potential biological risks and active behavioral factors that contribute to weight gain.
Beyond a single biomarker
The implications of this research go far beyond emotional hunger. Obesity is widely recognized to be a highly heterogeneous disease in which multiple biological pathways contribute to disease progression and treatment response.
For drug developers, this complexity poses significant challenges in target identification and patient stratification.
“This study opens a new door for identifying more complex targets,” O’Connor said. “By understanding which genetic pathways are activated in a particular phenotype, we can begin to point out which mechanisms are more responsive to treatment.”
Improving stratification in clinical trials
One of the most direct applications is in early stage research and clinical development. Obesity clinical trials often enroll broad patient populations, despite significant biological diversity among individuals. As a result, treatment responses can vary significantly.
By understanding which gene pathways are activated in a particular phenotype, we can begin to point out what mechanisms are driving responsiveness to treatment.
O’Connor believes that biological phenotyping could help address this challenge much earlier in the development process.
“This study allows for earlier and more accurate patient stratification by using biological signals, rather than relying solely on observable characteristics or self-reported symptoms,” he said.
“By proactively identifying patients who are more likely to suffer from emotional hunger, researchers can better define patient subgroups for clinical trials from the outset.”
Treatment response and patient matching
GLP-1 receptor agonists have transformed obesity treatment, but variable response and high discontinuation rates remain challenges.
Research suggests that patients with emotional hunger may respond differently to existing treatments, and that some patients may benefit more from behavioral interventions and treatments such as naltrexone/bupropion than from GLP-1 therapy alone.
O’Connor said the study also reflects the growing focus on precision obesity medicine.
“It is incredible to see innovative treatments being introduced in the obesity field now, and there will be many more on the way,” he said. “The real challenge is matching the right treatment to the right patient.”
O’Connor said this is the central purpose of the company’s MyPhenome test and related research efforts.
There are currently innovative treatments being introduced in the obesity field, and it’s amazing how many more are on the way.
“That’s what we’re accomplishing with Phenomics. Our MyPhenome test and ongoing research like this emotional hunger study will help identify distinct biological subtypes of obesity.”
By identifying the biological factors behind a patient’s condition, clinicians can make more informed treatment decisions and reduce reliance on trial-and-error prescriptions.
The impact may also extend to future drug development strategies.
“Looking to the future, these insights will have broader implications for drug discovery,” O’Connor explained. “By better defining patient populations, researchers can design and develop more targeted and effective treatments from the start.”
Ultimately, he believes these advances could help provide more personalized obesity care.
What comes next?
Future research will focus on validating the model in a clinical setting.
“The next step is to validate and refine predictors of emotional hunger in real-world settings,” O’Connor said.
Efforts include increasing clinical datasets and building partnerships with obesity drug manufacturers, payers, and health systems.
As researchers continue to explore more targeted approaches to obesity treatment, identifying biologically meaningful phenotypes such as emotional hunger may become increasingly important for both clinical care and drug development.
Although the research is still in its early stages, the findings demonstrate how biomarker strategies integrated with machine learning can help researchers better understand the biology of obesity and variation in treatment response.
