Thank you for your interest in our NIJ-funded study to establish a discipline-wide error rate estimate for the comparison of palm impressions (award # 2017-DN-BX-0170). We are pleased to report that we have completed data collection for this research and developed 2 online applications for exploring the results, which were presented at the 104th annual educational conference of the International Association for Identification in Reno, NV on August 15, 2019. If you would like to see the aggregate results, or your personal results (you will need your anonymous username), please visit https://cchampod.shinyapps.io/Results_BBStudy/. You can also explore the notion of frequentist confidence intervals and Bayesian credible intervals with the data from this study using the following link: https://cchampod.shinyapps.io/app_CI/. As a summary, 526 known ground truth cases of variable difficulty were created. 226 fingerprint examiners delivered a total of 12,279 suitability determinations and 9,460 comparison conclusions. Overall (removing inconclusive results), we report a false positive error rate of 0.67% and a false negative error rate of 9.46%.
If you would like to see the images associated with one of your trials, you can log in using your username and password from the study to the following URL: https://ips-labs.unil.ch/apps/pianos4-palmbb-nocnm. This version of PiAnoS will allow you to review your own annotations and see all mark images. You will only be able to see the exemplars for the same source trials because the exemplars for the different source trials were taken from a database search and we do not have consent to share those images in this format.
We will be submitting a manuscript that discusses the findings of this research to a peer-reviewed scientific journal later this year. Stay tuned to this page for updates on where you can find this as information becomes available. As always, we are thankful for your interest or participation. Please feel free to contact Heidi at heldridge@rti.org (directly) or LPResearch@rti.org (anonymously) if you have any questions.