Research
My work asks what a model can reliably infer from sensors that do not measure the target directly. I apply it to wearable health and to industrial control, and two supporting lines of work, explaining sensor data in language and running models on small devices, make those models easier to check and to deploy.
Application
Wearable health sensing
Estimating physiological state that a wearable cannot measure directly, with models a clinician can inspect.
With the Kireev Lab at UMass Amherst, I estimate blood pressure from graphene electronic tattoo patches with physics-informed and tree-based models, to track blood pressure continuously without a cuff. Our PhysioNet Challenge 2026 entry screens for future cognitive impairment from overnight sleep studies using interpretable sleep features aligned to the scored hospital. On the hidden test set it scored 0.678 age-conditioned AUROC, 10th of 42 teams. Earlier, I predicted hypoxemia severity in hospital in-patients from the photoplethysmograph.
- 2026
Cross-Site Generalization of Interpretable Sleep Features for Predicting Future Cognitive Impairment
S. A. I. Shouborno, Bashima Islam
Computing in Cardiology (CinC).
Accepted
Code BibTeX
@inproceedings{shouborno2026cinc,
title = {Cross-Site Generalization of Interpretable Sleep Features for Predicting Future Cognitive Impairment},
author = {Shouborno, S. A. I. and Islam, Bashima},
booktitle = {Computing in Cardiology (CinC)},
year = {2026}
}
- 2022
Res-SE-ConvNet: A Deep Neural Network for Hypoxemia Severity Prediction for Hospital In-Patients Using Photoplethysmograph Signal
T. I. Mahmud, S. A. Imran, C. Shahnaz
IEEE Journal of Translational Engineering in Health and Medicine, vol. 10, pp. 1–9, 2022.
PaperCode BibTeX
@article{mahmud2022resse,
title = {Res-SE-ConvNet: A Deep Neural Network for Hypoxemia Severity Prediction for Hospital In-Patients Using Photoplethysmograph Signal},
author = {Mahmud, T. I. and Imran, S. A. and Shahnaz, C.},
journal = {IEEE Journal of Translational Engineering in Health and Medicine},
volume = {10},
pages = {1--9},
year = {2022},
doi = {10.1109/JTEHM.2022.3217428}
}
Industrial cyber-physical systems
Finding faults in industrial control software from the plant's sensor data, without instrumenting the controller.
With Bosch Research, I reconstruct the branch-level program execution trace of an industrial controller from the plant's process sensors alone, so faults in control software can be localized without adding logging to the deployed controller. On a Bosch ctrlX testbed it detects code-path faults that sensor readings alone cannot reveal at 0.99 AUC, and with a healthy replay as reference it names the faulty function 69% of the time. During my Bosch internship I built a contrastive model linking four modalities from an industrial control testbed (vision, sensor, program trace and text data) for language-guided few-shot state classification.
- Under review
Program Execution Traces Without Instrumentation: Debugging Industrial Controllers from Physical Process Sensors
S. A. I. Shouborno, Emily Ruppel, Sudershan Boovaraghavan, Bashima Islam
AAAI Conference on Innovative Applications of Artificial Intelligence (IAAI-27).
Code BibTeX
@inproceedings{shouborno2027cpsdebugger,
title = {Program Execution Traces Without Instrumentation: Debugging Industrial Controllers from Physical Process Sensors},
author = {Shouborno, S. A. I. and Ruppel, Emily and Boovaraghavan, Sudershan and Islam, Bashima},
booktitle = {AAAI Conference on Innovative Applications of Artificial Intelligence (IAAI-27)},
year = {2027}
}
Supporting method
Explaining sensor data in language
Making a model's reading of sensor data inspectable by letting people ask about it in plain language.
LLaSA pairs an IMU encoder with a language model so it can answer open-ended questions about human activity from raw accelerometer and gyroscope data, and it outperforms GPT-3.5 and GPT-4o-mini on these tasks. We released the SensorCap and OpenSQA datasets, with 35,960 captions and 199,701 question and answer pairs. Concord studies when to trust LLM judges, using agreement among several judges as an uncertainty signal. This builds on two and a half years of production conversational AI and speech recognition for Bengali speakers.
- 2026
Concord: An Agreement-Aware Multi-Adjudication Pipeline for LLM Evaluation
Tyler Bliss, Mahit Verma, Aila Iyer-Singh, Subrata Biswas, Sheikh Asif Imran, Bashima Islam
Proceedings of the Fifth Workshop on Generation, Evaluation and Metrics (GEM), pp. 502–510, 2026.
Paper BibTeX
@inproceedings{bliss2026concord,
title = {Concord: An Agreement-Aware Multi-Adjudication Pipeline for LLM Evaluation},
author = {Bliss, Tyler and Verma, Mahit and Iyer-Singh, Aila and Biswas, Subrata and Imran, Sheikh Asif and Islam, Bashima},
booktitle = {Proceedings of the Fifth Workshop on Generation, Evaluation and Metrics (GEM)},
pages = {502--510},
year = {2026},
doi = {10.18653/v1/2026.gem-main.46}
}
- 2025
LLaSA: A Sensor-Aware LLM for Natural Language Reasoning of Human Activity from IMU Data
S. A. I. Shouborno, Mohammad Nur Hossain Khan, Subrata Biswas, Bashima Islam
Companion of the 2025 ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp Companion), pp. 893–899, 2025.
PaperarXivCode and datasets BibTeX
@inproceedings{shouborno2025llasa,
title = {LLaSA: A Sensor-Aware LLM for Natural Language Reasoning of Human Activity from IMU Data},
author = {Shouborno, S. A. I. and Khan, Mohammad Nur Hossain and Biswas, Subrata and Islam, Bashima},
booktitle = {Companion of the 2025 ACM International Joint Conference on Pervasive and Ubiquitous Computing (UbiComp Companion)},
pages = {893--899},
year = {2025},
doi = {10.1145/3714394.3756187}
}
Efficient inference on the device
Keeping these models within the battery and latency limits of wearables and embedded hardware.
JOLT trains early-exit networks so that each exit's confidence tracks which inputs it gets right, then routes on that confidence so that it overspends a stated compute budget by at most 6.8% when the input distribution shifts. Across seven sensor, image, audio and language datasets it ranks first of eleven methods. At WPI I was also a teaching assistant for the graduate course On-Device Deep Learning.
- 2026
JOLT: Jointly Optimized Layerwise Termination for Budgeted Early-Exit Inference
S. A. I. Shouborno, Mohammad Nur Hossain Khan, Bashima Islam
IEEE International Conference on Tools with Artificial Intelligence (ICTAI).
Accepted
Code BibTeX
@inproceedings{shouborno2026jolt,
title = {JOLT: Jointly Optimized Layerwise Termination for Budgeted Early-Exit Inference},
author = {Shouborno, S. A. I. and Khan, Mohammad Nur Hossain and Islam, Bashima},
booktitle = {IEEE International Conference on Tools with Artificial Intelligence (ICTAI)},
year = {2026}
}