Portinari : A Data Exploration Tool to Personalize Cervical Cancer - PowerPoint PPT Presentation

Portinari : A Data Exploration Tool to Personalize Cervical Cancer Screening Sagar Sen ∗ , Manoel Horta Ribeiro †, ∗ Certus V&V Center and HPV Research Group, 
 † Department of Computer Science, Universidade Raquel C. de Melo Minardi†, Federal de Minas Gerais, Belo Horizonte, Brazil, ‡ HPV Research Group, Cancer Registry of Norway, Wagner Meira Jr.†, Oslo, Norway Mari Nygård‡ 1

Software’s Impact in Societal Processes Through which means societies organize themselves? 2

Software’s Impact in Societal Processes Through which means societies organize themselves? politics 2

Software’s Impact in Societal Processes Through which means societies organize themselves? politics 2

Software’s Impact in Societal Processes Through which means societies organize themselves? politics social bonds 2

Software’s Impact in Societal Processes Through which means societies organize themselves? politics social bonds 2

Software’s Impact in Societal Processes Through which means societies organize themselves? politics social bonds labour 2

Software’s Impact in Societal Processes Through which means societies organize themselves? politics social bonds labour 2

Self-organizing vs. Centralized 3

Self-organizing vs. Centralized 3

Self-organizing vs. Centralized vs 3

Self-organizing vs. Centralized vs 3

4

Norway’s Screening Program AGUS/ASC-H/HSIL/ACIS/Ca Colposcopy Immediately with biopsy HPV Normal and Unsatisfactory HPV +ve LSIL or LSIL/AGUS/ 1 -3 months Immediately ASC-H/ACIS/Ca Immediately Entry intro screening program New ASC-US/LSIL HPV +ve 25-69 years of age Immediately 6 - 12 months Cytology Cytology HPV-test and HPV- Definitions test ACIS: Adenocarcinoma in situ HPV Normal Cytology AGUS: Atypical glandular cells of undetermined significance, 3 years Normal under cell. 3 years ASC-H: Irregular squamous cells with changes suggestive of high grade lesion, but do not meet all the criteria for diagnosis ASC-US: Irregular squamous cells with abnormalities of uncertain significance HPV Normal and Cytology Normal/ASC-US Ca: All types of cancer 3 years Cytology: Test to look at abnormal cells HPV: Humanpapillomavirus HSIL: High-grade squamous intraepithelial lesion LSIL: Low-grade squamous intraepithelial lesion 5

Triage Algorithm Invitation and Reminders Screening Exams 6

Triage Algorithm Invitation and Reminders Screening Exams 6

Verification and Validation Triage Algorithm Invitation and Reminders Screening Exams 6

Verification and Validation Triage Algorithm Invitation and Reminders Screening Exams 6

Verification and Validation Triage Algorithm Invitation and Reminders Screening Exams Statistics by epidemiologists 6

Verification and Validation Triage Algorithm Invitation and Reminders Screening Exams Statistics by epidemiologists Data exploration and visualization 6

7

What data? 8

9

What representation? 10

Querying 11

12

Portinari Flow Diagrams of Future Paths The flow diagram starts where the query patterns ends, Portinari allows two showing charts to be displayed side by side! Hovering over the edges or nodes of the graph shows stats! Each layer represents the distribution of diagnosis n exams into the future. 13

Explorations 14

Exploring effects of screening guidelines 3-4 years Cytology Normal Cytology Normal What happens after 3 years to this cohort? 15

Next Diagnosis in 3 years 230,331 individuals 261133 individuals 39 individuals 16

Next Two Diagnoses 193543 individuals 280572 individuals Did not return 17

Exploring existing screening guidelines 3-4 years Cytology Normal Cytology Normal What if women come back in the next 30 days? 18

Cytology 4337 not good individuals enough for some? 19

Comparing two scenarios in the triage algorithm next exam Cytology Normal HPV +ve Come back Come back in 1 to vs. within 1 year 2 years (Guideline) 20

Returned within 1 year Returned in 1 -2 years 13297 278 individuals individuals (227) 1.61% (19) HPV +ve 6.83% HPV +ve 21

Conclusion ➤ Alternative way of verifying and validating paths taken by people in a socio-technical system. ➤ Use of easy-to-query graph database. ➤ Generalizable to event-driven data. ➤ Generation hypothesis for further statistical investigation. 22

Questions? Manoel Horta Ribeiro https://github.com/manelhr/portinari/ manoelribeiro homepages.dcc.ufmg.br/~manoelribeiro/ 23

Questions? Manoel Horta Ribeiro https://github.com/manelhr/portinari/ manoelribeiro homepages.dcc.ufmg.br/~manoelribeiro/ 23