During my PhD, I worked on machine learning-driven decision support systems that bridge the gap between predictions and actions while operating in dynamic and uncertain environments. My research focused on applying techniques such as sequential decision-making, reinforcement learning, and federated learning, with applications spanning process analytics, healthcare, and operational decision-making.

Decision Support:

AI and ML have transformed how businesses and knowledge workers make decisions, both at operational and strategic levels. My PhD work leveraged deep learning architectures—LSTMs, MANNs, and Transformers—to model sequential decision processes and improve real-world decision-making during process execution.

  • Operational Decision Support: The flexible execution of business process instances entails multiple critical decisions involving various actors and objects, which can significantly impact process performance and the achievement of desired outcomes. These decisions, therefore, require careful attention, as suboptimal choices during process execution can lead to cost overruns, missed deadlines, and an increased risk of failure. In Project DeepProcess, we applied Memory-Augmented Neural Networks (MANNs) to recommend optimal actions to process users, enabling businesses to optimize process execution in real time. See [Paper] [Blog_post] for details.
  • Strategic Level: Project Alpha-GS explores game tree search with satisfiability (SAT) techniques to enable robust decision-making in adversarial settings. [Blog Post] [Paper]

Human-AI Collaboration in Healthcare

Building AI systems that work with humans—not replace them—is crucial. AI-driven decision support must integrate human expertise while maintaining interpretability, privacy, and reliability.

In one of my key projects, MIMIC-RL, I developed a Clinical Decision Support System for Sepsis Management using Deep Reinforcement Learning. The goal:

Given a patient's specific characteristics and physiological information at each time step as input, can an AI Agent learn an optimal treatment policy that prescribes the appropriate intervention (e.g., use of a ventilator) at each stage of the treatment process, in order to improve the final outcome (e.g., patient mortality)?

The framework models real-world patient trajectories to optimize medical interventions while preserving privacy and compliance. See [Blog_post] and [Paper] for details.

Federated Learning & Data Privacy

Privacy is a critical challenge in AI/ML research, particularly in sensitive domains like healthcare, where sensitive data is often distributed across multiple institutions and organizations. My research applied Federated Learning and Differential Privacy to mine and extract process models from decentralized datasets. Read the paper for more details.

The majority of human knowledge exists as unstructured text, and extracting actionable insights remains a challenge. Advances in large language models (LLMs) open new frontiers in human-AI interaction and knowledge retrieval.

  • Latent Semantic Indexing: Developed scalable semantic search techniques for retrieving relevant information from massive text corpora. [Blog Post]
  • Know-How Mining: Extracted process expertise from text using word2vec-based semantic similarity. [Paper] [Code]

Current Research Interests:

My recent focus has largely been on applied machine learning, even though i am been keeping an eye on progress in theoretical realm. Theoretical Machine learning is full of fascinating open problems like generalization, intelligent exploration vs exploitation, counter-factual evaluation, Meta-Learning and Sample Efficient Learning. Another challenge is that Decision-making systems need to be interpretable and explainable so that humans can understand and trust the actions suggested by AI models. Ensuring that the actions recommended by AI systems are fair and unbiased is crucial, especially in sensitive domains like healthcare, finance, and criminal justice. For effective decision-making, understanding the causal impact of actions is crucial. Currently, predictions made by many ML models are often based on correlations rather than causal relationships. Designing techniques and algorithms that can suggest actions based on predictions requires understanding how interventions will affect the outcome. This involves causal inference and counterfactual reasoning. Similarly, decision-making side of machine learning has been relatively neglected but its equally important. It involves individual high-stake decisions, decisions in the context of multiple decision makers, explanations for decisions and dialog about decisions.

Progress in solving these problems can have a major impact on application domains like industrial automation, healthcare, and education.

Publication List:

Some of the papers I’ve co-authored with awesome colleagues:

[1] Asjad Khan, Le H, Do K, Tran T, Ghose A, Dam H, Sindhgatta R. DeepProcess: Supporting Business Process Execution Using a MANN-Based Recommender System. In International Conference on Service-Oriented Computing 2021 Nov 22 (pp. 19-33). Springer

[2] Asjad Khan, Aditya Ghose, and Hoa Dam. Decision support for knowledge- intensive processes using RL-based recommendations. International Conference on Business Process Management. Springer, Cham, 2021.

[3] Asjad Khan, Aditya Ghose, and Hoa Dam. Cross-Silo Process Mining with Federated Learning. International Conference on Service-Oriented Computing. Springer, Cham, 2021.

[4] Santiputri M., N Deb, Aditya. K. Ghose, Hoa K. Dam, N. Chaki and Asjad Khan. Mining goal refinement patterns: Distilling know-how from data. In Proceedings of the 36th International Conference on Conceptual Modeling (ER-17), Springer Lecture Notes in Computer Science, Springer.

Thesis: A computational framework for Data-Driven Decision Support and Knowledge Augmented Process Analytics

Google Scholar Profile: Link

Great scientists tolerate ambiguity very well. They believe the theory enough to go ahead; they doubt it enough to notice the errors and faults so they can step forward and create the new replacement theory. If you believe too much you'll never notice the flaws; if you doubt too much you won't get started. It requires a lovely balance. - Richard Hamming