I work on efficient sequence modeling for healthcare AI, with a focus on Selective State Space Models (Mamba) as an alternative to Transformers for long clinical sequences.

Currently pursuing M.Sc. in AI & ML, with ongoing research collaboration at King Mongkut’s University of Technology Thonburi (KMUTT), Thailand.

My work is centered around real-world clinical constraints, not just model performance:

• Speech-based Cognitive Impairment Detection
  Modeling linguistic + acoustic signals for early detection of Mild Cognitive Impairment (MCI)

• Beyond Transformers
  Replacing attention-based architectures with Mamba (SSMs) for better scalability on long sequences

• Explainability in Clinical AI
  Making model decisions interpretable and clinically meaningful

• Cross-Dataset Generalization
  Robust modeling across datasets (e.g., Pitt → ADReSS → TAUKADIAL)

• Multimodal system combining speech transcription + sequence modeling
• Pipeline: Audio → Whisper → Text → Mamba → Clinical Prediction
• Focus on explainability + generalization
• Designed for real-world clinical deployment scenarios

Key Ideas:

• Replace Transformer-based encoders with Selective SSMs
• Integrate linguistic and interaction-level features
• Improve robustness across datasets

🔗 Code and paper links will be updated upon publication

• End-to-end pipeline for healthcare-oriented speech processing
• Includes:
  • Transcription (Whisper)
  • Feature extraction (linguistic + acoustic)
  • Sequence modeling (Mamba / DL models)

• Fusion Mamba for Explainable Speech-Based Detection of Mild Cognitive Impairment
  Under Review – JCSSE 2026

Core Areas:

• Sequence Modeling (SSMs, Deep Learning)
• Speech Processing & NLP
• Multimodal Learning

Core (What I actually use in research):

Working Knowledge:

Supporting Tools:

• Scaling Mamba-based architectures for long-sequence clinical data
• Integrating acoustic + linguistic multimodal fusion
• Exploring efficient alternatives to Transformer attention in healthcare AI

• LinkedIn: https://www.linkedin.com/in/santhanu-ajithkumar
• Email: santhanuusa@gmail.com