Ekaterina Kutafina (Dr. Dr. rer. medic.)

Sorting spikes from noisy single-channel in-vivo extracellular recordings is challenging, particularly due to the lack of ground truth data. Microneurography, an electrophysiological technique for studying peripheral sensory systems, employs experimental protocols that time-lock a subset of spikes. Stable propagation speed of nerve signals enables reliable sorting of these spikes. Leveraging this property, we established ground truth labels for data collected in two European laboratories and designed a proof-of-concept open-source pipeline to process data across diverse hardware and software systems. Using the labels derived from the time-locked spikes, we employed a supervised approach instead of the unsupervised methods typically used in spike sorting. We evaluated multiple low-dimensional representations of spikes and found that raw signal features outperformed more complex approaches, which are effective in brain recordings. However, the choice of the optimal features remained dataset-specific, influenced by the similarity of average spike shapes and the number of fibers contributing to the signal. Based on our findings, we recommend tailoring lightweight algorithms to individual recordings and assessing the “sortability feasibility” based on achieved accuracy and the research question before proceeding with sorting of non-time-locked spikes in future projects.

Large Language Models (LLMs) have been developed for a wide range of language-based tasks, while Reinforcement Learning (RL) has been primarily applied to decision-making problems such as robotics, game theory, and control systems. Nowadays, these two paradigms are integrated through different synergies. In this literature review, we focus on \textit{RL4LLM fine-tuning}, where RL techniques are systematically leveraged to fine-tune LLMs and align them with various preferences. Our review provides a comprehensive analysis of 230 recent publications, presenting a methodological taxonomy that organizes current research into three primary method domains: \textit{Optimization Algorithm}, concerning innovation in core RL update rules; \textit{Training Framework}, regarding innovation in the orchestration of the training process; and \textit{Reward Modeling}, addressing how LLMs learn and represent preferences and feedback. Within these primary domains, we further analyze methods and innovations through more granular categories to provide an in-depth summary of RL4LLM fine-tuning research. We address three research questions: 1) recent methods overview, 2) methodological innovations, and 3) limitations and future directions. Our analysis comprehensively demonstrates the breadth and impact of recent RL4LLM fine-tuning research while highlighting valuable directions for future investigation.

Cervical cancer is the third most common cancer in women, and recent studies have highlighted the importance of body composition markers in predicting patient outcomes. We build upon the data of 83 patients from the Uterus-11 study, to explore the relation of pairwise feature combinations to long-term progression-free survival. We propose a framework to identify the parameter combinations with pre-defined thresholds of “normal range” which provide good separation of the survival group. Further, we optimize the pair-wise thresholds to further improve the separation measured by F1 scores. This approach allowed us to improve the statistical significance of hazard ratios in comparison to the previous studies. The optimization results suggest that the normal ranges of well-established biomarkers such as body mass index could be shifted in the context of specific diseases to achieve optimal outcome.

Introduction:

The Local Data Hub (LDH) is a platform for FAIR sharing of medical research (meta-)data. In order to promote the usage of LDH in different research communities, it is important to understand the domain-specific needs, solutions currently used for data organization and provide support for seamless uploads to a LDH. In this work, we analyze the use case of microneurography, which is an electrophysiological technique for analyzing neural activity.

Methods:

After performing a requirements analysis in dialogue with microneurography researchers, we propose a concept-mapping and a workflow, for the researchers to transform and upload their metadata. Further, we implemented a semi-automatic upload extension to odMLtables, a template-based tool for handling metadata in the electrophysiological community.

Results:

The open-source implementation enables the odML-to-LDH concept mapping, allows data anonymization from within the tool and the creation of custom-made summaries on the underlying data sets.

Discussion:

This concludes a first step towards integrating improved FAIR processes into the research laboratory’s daily workflow. In future work, we will extend this approach to other use cases to disseminate the usage of LDHs in a larger research community.

This work aims to improve FAIR-ness of the microneurography research by integrating the local (meta)data to existing research data infrastructures. In the previous work, we developed an odML based solution for local metadata storage of microneurography data. However, this solution is limited to a narrow community. As a next step, we propose the integration into the Local Data Hubs, data-sharing services within NFDI4Health infrastructure. We outline a first concept, that streams chosen data from the established odMLtables GUI.