Characterizing of adverse events across 30,000 peer reviewed case report publications.
Bibliographic record
Abstract
e18858 Background: Pharmacovigilance suffers from considerable underreporting. This is particularly the case for novel oncology therapeutics where regulatory approval is supported by clinical trials with small sample sizes and limited follow-up to observe later treatment-related adverse events. More recently approval of oncologic therapies based on Real World Evidence, using study designs such as historical controls further complicates the task of assessing the clinical significance of AE. Furthermore, causation is rarely confirmed outside controlled settings such as clinical trials, and particularly absent in conventional RWE sources. Methods: We evaluated the ability of published case reports to serve as a source of pharmacovigilance for novel oncology therapeutics as well as the prevalence of reporting of clinician confirmed causality. We assessed structured patient journey data extracted from approximately 30,000 peer reviewed oncology case reports published since 2015 and available from the OpenCaseTM database. The reviewed cases covered hematological cancers (myelomas, leukemias, and lymphomas) and solid cancers (lung, breast, colorectal, and bladder) and represented case reports from over 125 nations. Adverse events and key patient journey data were extracted using state-of-the-art natural language processing (NLP) models with validated accuracy of over 92%. Results: A total of 58.5% reported at least one adverse event after or during a treatment period. Of these, the median number of reported adverse events were 4 (IQR: 2 - 8). There were 12,551 unique AEs after normalizing with the MedDRA database using a state-of-the-art AI model. The ten most frequently reported AEs were pain, fever, weight loss, fatigue, post-op complications, fatigue, anemia, infection, bleeding, and dyspnea, ranging from 897 to 2606 occurrences. Clinical confirmation of causation (i.e., treatment-related adverse event) was available for at least one adverse event in over half of the case reports. The majority of these were related to immune checkpoint inhibitors, but toxicity due to chemotherapies and complications due to surgery were also common. Conclusions: In conclusion, our study suggests that peer reviewed oncology case reports represent a rich source of reported adverse events, often with clinically confirmed causality, which could substantially augment adverse events reporting of oncology therapeutics.
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 | 0.028 | 0.148 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.035 | 0.025 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.012 | 0.004 |
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".