[Paper Review] Machine Learning-Based Classification of Jhana Advanced Concentrative Absorption Meditation (ACAM-J) using 7T fMRI

Puneet Kumar, Winson F. Z. Yang+3 more|arXiv (Cornell University)|Feb 13, 2026

Mindfulness and Compassion Interventions0 citations

TL;DR

The paper uses fMRI-derived regional homogeneity (ReHo) features from 7T fMRI to train machine-learning classifiers that distinguish ACAM-J meditation from non-meditative states, achieving about 66.82% accuracy with feature-importance highlighting prefrontal and anterior cingulate involvement.

ABSTRACT

Jhana advanced concentration absorption meditation (ACAM-J) is related to profound changes in consciousness and cognitive processing, making the study of their neural correlates vital for insights into consciousness and well-being. This study evaluates whether functional MRI-derived regional homogeneity (ReHo) can be used to classify ACAM-J using machine-learning approaches. We collected group-level fMRI data from 20 advanced meditators to train the classifiers, and intensive single-case data from an advanced practitioner performing ACAM-J and control tasks to evaluate generalization. ReHo maps were computed, and features were extracted from predefined brain regions of interest. We trained multiple machine learning classifiers using stratified cross-validation to evaluate whether ReHo patterns distinguish ACAM-J from non-meditative states. Ensemble models achieved 66.82% (p < 0.05) accuracy in distinguishing ACAM-J from control conditions. Feature-importance analysis indicated that prefrontal and anterior cingulate areas contributed most to model decisions, aligning with established involvement of these regions in attentional regulation and metacognitive processes. Moreover, moderate agreement reflected in Cohen's kappa supports the feasibility of using machine learning to distinguish ACAM-J from non-meditative states. These findings advocate machine-learning's feasibility in classifying advanced meditation states, future research on neuromodulation and mechanistic models of advanced meditation.

Motivation & Objective

• Investigate neural correlates of ACAM-J meditation using high-field fMRI data.
• Evaluate whether ReHo patterns across predefined brain regions can distinguish ACAM-J from non-meditative states.
• Assess the feasibility of machine learning for classifying advanced meditation states.
• Identify brain regions most influential in classifier decisions related to attention regulation and metacognition.

Proposed method

• Compute fMRI regional homogeneity (ReHo) maps for participants.
• Extract features from predefined brain regions of interest.
• Train multiple machine-learning classifiers with stratified cross-validation.
• Evaluate classifier performance in distinguishing ACAM-J from control conditions.
• Perform feature-importance analysis to identify influential brain regions.

Experimental results

Research questions

• RQ1 Can ReHo-based features from 7T fMRI distinguish ACAM-J meditation from non-meditative states using machine learning?
• RQ2 Which brain regions contribute most to the classification of ACAM-J state?
• RQ3 What is the generalizability of a model trained on group data to a single practitioner performing ACAM-J?
• RQ4 Is there statistically significant accuracy in differentiating ACAM-J from control conditions?

Key findings

• Ensemble models achieve 66.82% accuracy in distinguishing ACAM-J from control conditions (p < 0.05).
• Feature-importance analysis implicates prefrontal and anterior cingulate regions as top contributors.
• Findings align with known roles of attention regulation and metacognitive processing in these regions.
• Moderate agreement observed via Cohen’s kappa supports feasibility of ML-based state classification.
• Results support future work on neuromodulation and mechanistic models of advanced meditation.