Developing clinical ML systems is costly and labor-intensive due to fragmented preprocessing, privacy constraints, and model-data alignment challenges. We introduce a modular agentic AI framework that automates the end-to-end ML lifecycle, from ingestion and anonymization to preprocessing, model selection, and interpretable inference. Each agent performs a well-defined task, enabling scalable workflows across structured and unstructured data. We evaluate the framework on public datasets from geriatrics, palliative care, and colonoscopy imaging. Data are automatically classified, anonymized via DLP, semantically represented, and mapped to suitable models using embedding- or LLM-based strategies. Preprocessing and inference agents ensure compatibility and produce interpretable outputs (e.g., SHAP, attention maps). By consolidating manual tasks into coordinated autonomous agents, our approach reduces expert intervention, lowers operational costs, and supports scalable clinical ML deployment.

Location-based distributed communication in underground mines has been a hard problem to solve due to unreliable centralized architecture such as leaky feeder systems, high attenuation, and the unavailability of GPS signals. Delay Tolerant Networks (DTN) enable decentralized message routing using the store-carry-forward method that can help in creating situational awareness needed to handle emergency and disaster scenarios. The ability to predict where the DTN nodes (miner) might have been at/are headed to (with respect to the mine regions and pillars) at different times, combined with contact-based routing and intelligent handling of buffer, can be used for better delivery of messages. To this end, we propose a hybrid approach, called MinerRouter, that uses Random Forest (RF) and Graph Autoencoder (GAE) - Long Short Term Memory (LSTM) model to exploit the short- and long-mobility patterns of miners, respectively for faster message/content dissemination. Our simulations show that MinerRouter outperforms Opportunistic RF (RF), Opportunistic Contact Graph Routing (O-CGR), MaxProp, SemiBlind, and Blind routing protocols in terms of the delivery ratio of messages received, message latency, buffer occupancy Rate, communication overhead costs, and hop count.

Recent reports by the Mine Safety and Health Administration suggest that several injuries and fatalities could be attributed to the inability to accurately locate miners in case of disasters. Since underground mines have a complicated geometrical landscape and technological constraints such as no GPS information available, it is difficult to predict the location of a miner and hence may cause delays and inefficiencies in rescue operations during a disaster. A significant amount of research has been done to capture complex spatio-temporal relationships of movement of the nodes/people/things with time, spatial and temporal features to separately extract these relationships for location prediction. Although Markov Chains (MC) and Recurrent Neural Network (RNN) based methods have been used to predict locations, not all of them specifically mention the spatial locations, their connections and the aggregation techniques which would allow for the actual representations of the trajectory of miners. Addressing these concerns, we develop a first-of-its-kind end-to-end system entitled MinerFinder to predict the future location of the miners by incorporating Long Short Term Memory (LSTM) for trajectory information with Graph Autoencoder (GAE) for spatial environmental information representing the node connectivity. In addition, our approach will combine the miners' previous trajectories and daily repetitive patterns enhancing the prediction robustness. We evaluated MinerFinder over synthetic dataset to analyze the structure and location topology of an underground mine compared with foreground locations. Our model outperforms state of the art models and achieves an AP score ranging from (0.62 - 0.68) and Receiver Operating Characteristics (ROC) ranging from (0.63--0.68) with increasing percentage of prominent locations (most visited) to 50%.