Extracting human data from published figures: implications for data science and bioethics
Bibliographic record
Abstract
Abstract The advent of text mining and natural text reading artificial intelligence has opened new research opportunities on the large collections of research publications available through journal and other resources. These systems have begun to identify novel connections or hypotheses due to an ability to read and extract information from more literature than a single individual could in their lifetime. Most research publications contain figures where data is represented in a graph. Modern publication guidelines are strongly encouraging publication of graphs where all data is displayed as apposed to summary figures such as bar charts. Figures are often encoded in a graphing language that is interpreted and displayed as a graphics. Conversion figures in publications to the underlying code should enable text-based mining to extract the underlying raw data of the graph. Here I show that data from publications greater than 15 years old that contain time series data on human patients is extractable from the original publication and can be reassessed using modern tools. This could benefit cases where data sets are not available due to file loss or corruption. This may also create and issue for the publication of human data as sharing of human data often requires research ethics approval. Author summary Figures embedded in published research manuscripts are a minable resource similar to text mining. Figures are text based code that draws the image, as such the underlying text of the code can be used to reassemble the original data set.
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.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.101 | 0.516 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.010 | 0.018 |
| Science and technology studies | 0.003 | 0.009 |
| Scholarly communication | 0.022 | 0.017 |
| Open science | 0.004 | 0.007 |
| Research integrity | 0.004 | 0.006 |
| Insufficient payload (model declined to judge) | 0.018 | 0.011 |
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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".