Different Convergence Parameters Applied to the S-PLUS GAM Function
Bibliographic record
Abstract
ArticlePlus Click on the links below to access all the ArticlePlus for this article. Please note that ArticlePlus files may launch a viewer application outside of your web browser. https://links.lww.com/EDE/A34 https://links.lww.com/EDE/A35 To the Editor: Recently the Health Effects Institute (HEI) circulated a letter from the investigators at the Johns Hopkins Bloomberg School of Public Health who conducted the National Morbidity, Mortality and Air Pollution Study (NMMAPS) on short-term effects of air pollution on health, in which they report a problem with the default convergence criteria in the S-PLUS GAM function they have been using. Applying more stringent convergence criteria resulted in reducing the mortality increase associated with an increase of 10 μg/m3 in PM10 concentrations from 0.44% to 0.22%. In this report questions were raised about the effect of this problem on the results of other similar studies. The European multicenter project Air Pollution and Health: A European Approach (APHEA), with objectives similar to the NMMAPS, has published results on particulate matter effects on total mortality in a recent issue of Epidemiology, 1 using the S-PLUS GAM function with default convergence criteria for some of the analyses. We have reanalyzed our data using more stringent convergence criteria as proposed by the Johns Hopkins Group (specifically, the max number of iterations was set to 1,000 and the difference of two successive coefficients to 10−14). We found that the change in the estimated effect was marginal. The estimated combined increases in mortality associated with an increase in 24-hour PM10 (particulate matter less than 10 μm in aerodynamic diameter) and black smoke (BS) concentrations by + 10 μg/m3 were reduced by 4% and remain identical when reported with one significant digit; for PM10 the increase in mortality was 0.6% (95% confidence interval = 0.4%–0.8%) and for BS the increase was 0.6% (0.3%–0.8%). Tables showing individual city results using both the default and the more stringent convergence parameters are available with the electronic version of this letter at http://www.epidem.com. It appears, therefore, that the extent of the bias reported in the NMMAPS is not necessarily applicable in all studies that use the GAM function. The reasons may lie in differences concerning the smooth functions introduced in the model, the number of the degrees of freedom, the lag times considered, the way by which confounders have been adjusted, or the data patterns. Small changes in the model-derived estimates are expected under any change in the modeling procedure and are, in fact, trivial in comparison with differences among individual city estimates within each project. Optimization of model choice is a continuous procedure and the initiated collaboration between NMMAPS, APHEA and Canadian researchers will soon address further methodologic issues. Klea Katsouyanni Giota Touloumi Evangelia Samoli Alexandros Gryparis Yannis Monopolis Alain LeTertre Azedine Boumghar Giuseppe Rossi Denis Zmirou Ferran Ballester Hugh Ross Anderson Bogdan Wojtyniak Anna Paldy Rony Braunstein Juha Pekkanen Christian Schindler Joel Schwartz
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 | 0.001 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.001 | 0.000 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.007 | 0.012 |
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; both teacher heads agree on what is shown here.
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".