Siva

Professor @ Bioengineering, CMC Vellore

[Siva @siva82kb]

When you draw or write, your hand slows at curves and speeds up on straight paths — the speed–curvature power law.

Biology or mathematical illusion?

My latest post digs in, with real data and an interactive visualization.

https://t.co/VJOmW1Oyuc

#repairingbrokenmovements

[Siva @siva82kb]

Covers: why mean jerk fails, how dimensionless jerk fixes it, the dynamic range problem, and LDLJ's practical limitations — noise sensitivity, non-monotonicity, and amplitude threshold bias.

With 5 interactive simulations.

#repairingbrokenmovements #neurorehab #kinometrics

[Siva @siva82kb]

New post on the Log Dimensionless Jerk (LDLJ) measure for quantifying movement smoothness.

In this new post I walk through the jerk-based approach — from its roots in the minimum jerk model of human movement.

👉 https://t.co/F1MvQPPHdz

[Siva @siva82kb]

SPARC measures the "wiggliness" of the Fourier magnitude spectrum — and the reason it gets wiggly is the same physics behind the double-slit experiment in optics.

New post with 5 interactive simulations: https://t.co/akmRJTIru4

#repairingbrokenmovements #neurorehab #kinometrics

[Siva @siva82kb]

It's been >10 years since we published the SPARC measure of movement smoothness. Misconceptions about how it works persist.

Most people think SPARC quantifies the amount of high-frequency components in a movement. It does not.

[Siva @siva82kb]

Copernicus put us among the rocks;
Darwin put us among the animals;
AI put us among the calculators.
— a thought I've been sitting with
#AI #Philosophy #Science

[Siva @siva82kb]

Congratulations to the whole team 👏!

We look forward to the impact this innovation will have on neurorehab in India and abroad.

#NeuroRehab #MedTech #CMCVellore #StrokeRehabilitation

[Siva @siva82kb]

Exciting news from @CMCVellore 🎉
https://t.co/fRYZrx4Cs5

80% of stroke survivors in India have upper limb impairments. Quality neurorehab shouldn't be limited to urban centres.

[Siva @siva82kb]

Started as an MS thesis by Sujith Christopher (Dept. of Bioengineering, CMC Vellore). Now licensed to Dhariki Labs Pvt. Ltd. — Technology Transfer Agreement signed 5 Feb 2026.

[Siva @siva82kb]

We hope the causal model approach is a fruitful direction to pursue for improved individual patient outcomes.
Getting to a fully specified, practically useful causal model is going to take some work. But, I am excited!

#neurorehab #rehabrobots #causaldag

[Siva @siva82kb]

Preprint alert! 🚨

Our new preprint on a mechanistic understanding of robot‑assisted upper‑limb therapy is now online at https://t.co/k6s1U7Smgv:
https://t.co/JuD4ynDgwH

[Siva @siva82kb]

We then outline phenomena the model can already explain, its testable predictions, its limitations, and how such a model can support the broader vision of precision neurorehab.

[Kiran Mazumdar-Shaw @kiranshaw]

I had an overseas business visitor to Biocon Park who said ‘ Why are the roads so bad and why is there so much garbage around? Doesn’t the Govt want to support investment? I have just come from China and cant understand why India can’t get its act together especially when the winds are favourable?’ @siddaramaiah @DKShivakumar @PriyankKharge

[Siva @siva82kb]

Such a dataset could help identify computationally efficient solutions that perform consistently across the wider patient population/sub-populations, while minimizing or eliminating the requirement for annotated subject-level data.
#neurorehab #ulfunctioning #uluse #autoencoders

[Siva @siva82kb]

https://t.co/VGCZrBtZdL

Continuing our exploration of methods to quantify upper-limb (UL) use, here is our work on autoencoders led by Parvathy — a joint PhD student with Prof. Varadhan SKM (IITMadras) & myself.

[Siva @siva82kb]

These findings are based on a small dataset of 10 healthy and 5 hemiparetric subjects.

There is an urgent need to validate existing UL use detection methods using a large, diverse dataset.