876
Views
0
CrossRef citations to date
0
Altmetric
Research Article

Explainable Analytics to Predict the Quality of Life in Patients with Prostate Cancer from Longitudinal Data

ORCID Icon, ORCID Icon, ORCID Icon & ORCID Icon
Article: 2055393 | Received 16 Mar 2021, Accepted 16 Mar 2022, Published online: 15 Apr 2022

Figures & data

Table 1. Summary of papers describing predicting tools relating to pros (September 2016 – July 2019). ng (not given), pros (patient-reported outcomes).

Table 2. Explanations of the models for predictive tools in pros. ng (not given), pros (patient-reported outcomes).

Table 3. Statistical characteristics of patients for all the epic-26 domains score quartiles in 1-year follow-up. epic-26 domains are UI, UIO, b (bowel), s (sexual) and h (hormonal).

Table 4. Top 10 important variables from baseline visit selected by CB-DNN for predicting 1-year follow-up UI along with their definitions. .

Table 5. Top 10 important variables from baseline visit selected by CB-DNN for predicting 1-year follow-up UIO along with their definitions. .

Table 6. Top 10 important variables from baseline visit selected by CB-DNN for predicting 1-year follow-up bowel along with their definitions.

Table 7. Top 10 important variables from baseline visit selected by CB-DNN for predicting 1-year follow-up sexual along with their definitions.

Table 8. Top 10 important variables from baseline visit selected by CB-DNN for predicting 1-year follow-up hormonal along with their definitions.

Figure 1. R-squared between predicted and actual values in EPIC-26 domains. EPIC-26 Domains are a) UI b) UIO c) Bowel d) sexual and e) Hormonal.

Figure 1. R-squared between predicted and actual values in EPIC-26 domains. EPIC-26 Domains are a) UI b) UIO c) Bowel d) sexual and e) Hormonal.

Table 9. The items related to each EPIC-26 domain. EPIC-26 Domains are UI, UIO, B (Bowel) and S (sexual).

Figure 2. BNs explanation for UI. eq_004 (Usual activities?) has the following options:1, I have no problems doing my usual activities | 2, I have slight problems doing my usual activities | 3, I have moderate problems doing my usual activities | 4, I have severe problems doing my usual activities | 5, I am unable to doing my usual activities. im_001 (Bone scan?) has the following options: 1, yes | 2, No.

Figure 2. BNs explanation for UI. eq_004 (Usual activities?) has the following options:1, I have no problems doing my usual activities | 2, I have slight problems doing my usual activities | 3, I have moderate problems doing my usual activities | 4, I have severe problems doing my usual activities | 5, I am unable to doing my usual activities. im_001 (Bone scan?) has the following options: 1, yes | 2, No.

Figure 3. BNs explanation for UIO. ep_015 (Abdominal/Pelvic/Rectal pain?) has the following options: 1, No problem | 2, Very small problem | 3, Small problem | 4, Moderate problem | 5, Big problem. ib_064 (% of first RMZ core involved?) is a numerical variable which those less than the mean and greater than the mean are labeled with 0 and 1 respectively.

Figure 3. BNs explanation for UIO. ep_015 (Abdominal/Pelvic/Rectal pain?) has the following options: 1, No problem | 2, Very small problem | 3, Small problem | 4, Moderate problem | 5, Big problem. ib_064 (% of first RMZ core involved?) is a numerical variable which those less than the mean and greater than the mean are labeled with 0 and 1 respectively.

Figure 4. BNs explanation for Bowel (B). ib_042 (Number of positive cores biopsied from LMZ?) has the following options: 1, 1 | 2, 2 | 3, Fragmented core tissue. ep_012 (Increased frequency of bowel movements?) has the following options: 1, No problem | 2, Very small problem | 3, Small problem | 4, Moderate problem | 5, Big problem.

Figure 4. BNs explanation for Bowel (B). ib_042 (Number of positive cores biopsied from LMZ?) has the following options: 1, 1 | 2, 2 | 3, Fragmented core tissue. ep_012 (Increased frequency of bowel movements?) has the following options: 1, No problem | 2, Very small problem | 3, Small problem | 4, Moderate problem | 5, Big problem.

Figure 5. BNs explanation for Sexual (S). ep_017 (How would you describe your ability to have an erection during the last 4 weeks?) has the following options: 1, Very poor to none | 2, Poor | 3, Fair | 4, Good | 5, Very good. ep_018 (How would you describe your ability to reach orgasm (climax) during the last 4 weeks?) has the following options: 1, Very poor to none | 2, Poor | 3, Fair | 4, Good | 5, Very good.

Figure 5. BNs explanation for Sexual (S). ep_017 (How would you describe your ability to have an erection during the last 4 weeks?) has the following options: 1, Very poor to none | 2, Poor | 3, Fair | 4, Good | 5, Very good. ep_018 (How would you describe your ability to reach orgasm (climax) during the last 4 weeks?) has the following options: 1, Very poor to none | 2, Poor | 3, Fair | 4, Good | 5, Very good.

Figure 6. BNs explanation for Hormonal (H). eq_002 (Mobility?) has the following options: 1, I have no problems in walking about | 2, I have slight problems in walking about | 3, I have moderate problems in walking about | 4, I have severe problems in walking about | 5, I am unable to walk about. ib_008 (Gland volume?) is a numerical variable which those less than the mean and greater than the mean are labeled with 0 and 1 respectively.

Figure 6. BNs explanation for Hormonal (H). eq_002 (Mobility?) has the following options: 1, I have no problems in walking about | 2, I have slight problems in walking about | 3, I have moderate problems in walking about | 4, I have severe problems in walking about | 5, I am unable to walk about. ib_008 (Gland volume?) is a numerical variable which those less than the mean and greater than the mean are labeled with 0 and 1 respectively.

Table 10. What-If Analysis Scenarios for Five Patients.

Table 11. Variable importance ranking comparisons between global algorithm CB-DNN and local BNs.