Automatic Identification of Information Quality Metrics in Health News Stories
Bibliographic record
Abstract
Objective: Many online and printed media publish health news of questionable trustworthiness and it may be difficult for laypersons to determine the information quality of such articles. The purpose of this work was to propose a methodology for the automatic assessment of the quality of health-related news stories using natural language processing and machine learning. Materials and Methods: We used a database from the website HealthNewsReview.org that aims to improve the public dialogue about health care. HealthNewsReview.org developed a set of criteria to critically analyze health care interventions' claims. In this work, we attempt to automate the evaluation process by identifying the indicators of those criteria using natural language processing-based machine learning on a corpus of more than 1,300 news stories. We explored features ranging from simple n-grams to more advanced linguistic features and optimized the feature selection for each task. Additionally, we experimented with the use of pre-trained natural language model BERT. Results: For some criteria, such as mention of costs, benefits, harms, and “disease-mongering,” the evaluation results were promising with an F 1 measure reaching 81.94%, while for others the results were less satisfactory due to the dataset size, the need of external knowledge, or the subjectivity in the evaluation process. Conclusion: These used criteria are more challenging than those addressed by previous work, and our aim was to investigate how much more difficult the machine learning task was, and how and why it varied between criteria. For some criteria, the obtained results were promising; however, automated evaluation of the other criteria may not yet replace the manual evaluation process where human experts interpret text senses and make use of external knowledge in their assessment.
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 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 | 0.002 | 0.002 |
| 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 | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 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 itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, 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".