Alex Dhima's Avatar

Alex Dhima

@alexdhima.bsky.social

Postbac IRTA @NIMH Section on Development and Affective Neuroscience | machine learning๐Ÿค– + digital tools๐Ÿ“ฑ + neuroscience ๐Ÿง  to personalize mental health | formerly @HarvardMedical & @UMassAmherst

43 Followers  |  67 Following  |  11 Posts  |  Joined: 12.12.2024
Posts Following

Posts by Alex Dhima (@alexdhima.bsky.social)

Preview
Research Assistant I Company Description: By working at Harvard University, you join a vibrant community that advances Harvard's world-changing mission in meaningful ways, inspires innovation and collaboration, and builds...

We're hiring! The Emotion and Psychopathology in Context Lab at Harvard invites applications for a full-time RA/lab manager to start in 2026 summer. Ideal for students preparing for graduate study (e.g., clinical psych). Please consider sharing & applying! jobs.smartrecruiters.com/HarvardUnive...

24.02.2026 17:38 โ€” ๐Ÿ‘ 31    ๐Ÿ” 28    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 1

Congrats!!!

24.02.2026 02:50 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

Thanks so much Grace ๐Ÿ˜Š

02.01.2026 16:33 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 0    ๐Ÿ“Œ 0
OSF

A huge thanks to Manasa Kalanadhabhatta and Adam Grabell their mentorship and support on this project!

View full preprint ๐Ÿ‘‰ doi.org/10.31234/osf...

31.12.2025 08:26 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

โ€ฆ this approach may hold promise for personalized, neuroscience-informed treatment routing and response monitoring in early childhood.

31.12.2025 08:26 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

We show that naturalistic, brain-based measures via fNIRS, analyzed using simple and interpretable machine learning models, can detect psychotherapy-like states in young children, although modestly. While larger replications with external datasets are required โ€ฆ

31.12.2025 08:26 โ€” ๐Ÿ‘ 1    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

๐Ÿ“ŠFeature analysis revealed that lower mean activity and variability in the left ventrolateral PFC while engaging in calm-emotion๐Ÿ˜Œ reflection increased the likelihood of being classified into the therapy-like condition.

31.12.2025 08:26 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

โ€ฆ whereas post-hoc exploratory models (using the four most frequently selected features across 1,000 folds from the a-priori models) yielded improved and consistently significant performance (mean AUC=0.72 - 0.76, p=0.005) ๐ŸŽฏ

31.12.2025 08:26 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

๐Ÿ“ŠAcross 200 iterations of 5-fold cross-validation, logistic regression models accurately classified whether a child was engaged in the therapy-like task (mean AUC=0.60-0.76). Performance in a-priori models was modest against a permutation null (AUC=0.60-0.63, p=0.044-0.13)โ€ฆ

31.12.2025 08:26 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0

๐Ÿง fNIRS was used to record activity of the bilateral dorsolateral and ventrolateral prefrontal cortex (PFC) throughout the entire duration of the task.

31.12.2025 08:26 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
Post image

78 preschool children were randomly assigned to:
- a dyadic coloring task prompting awareness and reflection of ๐Ÿ˜กangry and๐Ÿ˜Œcalm emotions, designed to mimic a therapy-like interaction, or
- identically structured interpersonal control condition with no emotion-related prompts๐Ÿšซ

31.12.2025 08:26 โ€” ๐Ÿ‘ 0    ๐Ÿ” 0    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0
OSF

New preprintโ€ผ๏ธ

Thrilled to end 2025 with new work extending my undergrad honors thesis showing that interpretable machine learning can classify young childrenโ€™s engagement in dyadic, psychotherapy-like tasks from live fNIRS data of the vlPFC and dlPFC ๐Ÿง 

๐Ÿงต๐Ÿ‘‡

OSF๐Ÿ‘‰ doi.org/10.31234/osf...

31.12.2025 08:26 โ€” ๐Ÿ‘ 13    ๐Ÿ” 3    ๐Ÿ’ฌ 1    ๐Ÿ“Œ 0