A Conversational Large‐Language‐Model Tutor that Accelerates Machine‐Learning Method Development in Routine Bioanalytical Workflows
Why this work is in the frame
A frame that forgets how it found something cannot be audited. These are the routes that admitted this work.
Bibliographic record
Abstract
As machine learning (ML) becomes increasingly relevant in experimental chemistry, many scientists face barriers to adoption due to limited training in ML. While AutoML platforms offer powerful capabilities, they lack the instructional scaffolding needed by users without an ML background. To address this gap, a lightweight, conversational assistant is presented that guides users through ML workflow design using plain-language dialog. Powered by OpenAI's GPT-4o and deployed via a Gradio interface, the assistant operates under a structured system prompt that simulates pedagogical reasoning. It behaves like a domain-specific tutor: helping users define ML goals, assess data structure, select models, evaluate metrics, and generate annotated Python code. A complete documentation of the development process is provided, allowing researchers to adapt the system for other domains. Herein, its utility is demonstrated in two representative case studies: 1) image classification of lateral flow immunoassay test strips for diagnostic readout; and 2) regression-based prediction of liquid chromatography-mass spectrometry retention times from molecular descriptors for small molecules. In both cases, lab members with no ML experience successfully developed working models guided solely by the assistant. By lowering the barrier to ML adoption in data-rich analytical workflows, this system offers a customizable workflow for building domain-specific assistants across experimental science.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.003 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it