bild 1home.thep.lu.se/~anders/atp_slides/140205-mattias.pdftitle: bild 1 author: mattias created...
TRANSCRIPT
![Page 1: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/1.jpg)
Astronomy & Theoretical Physics Lund University
Modeling of survival data and some nice applications in clinical medicine
Mattias Ohlsson
![Page 2: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/2.jpg)
Astronomy & Theoretical Physics Lund University
Overview of activities
Machine learning
development
Heart surgery
Breast cancer
Myocardial perfusion
scintigraphy
Cell biology
Acute coronary syndromes
Alzheimer's diseaseProstate cancer
Bone Scan Index
Exini Diagnostics AB
![Page 3: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/3.jpg)
Astronomy & Theoretical Physics Lund University
Common theme: survival analysis
Machine learning
development
Heart surgery
Prostate cancer
Breast cancer
Bone Scan Index
![Page 4: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/4.jpg)
Astronomy & Theoretical Physics Lund University
Survival analysis ~ adding a temporal information to pattern recognition problems
“Classical” pattern recognition
![Page 5: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/5.jpg)
Astronomy & Theoretical Physics Lund University
Survival analysis ~ analysis of time duration until an event occurs
Time
E.g.Therapy,
Transplantation
Pattern Event
E.g.Relapse of cancer,
Death
![Page 6: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/6.jpg)
Astronomy & Theoretical Physics Lund University
Survival analysis – the data
Time
Pattern information
Event
Censored event,eg. still alive, leftthe study, death because of otherreasons etc
![Page 7: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/7.jpg)
Astronomy & Theoretical Physics Lund University
Survival analysis – what do we want to model?
The survival function
![Page 8: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/8.jpg)
Astronomy & Theoretical Physics Lund University
The hazard function
(event rate at time t conditional on survival until time t or later.Interpretation: risk of dying at time t)
![Page 9: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/9.jpg)
Astronomy & Theoretical Physics Lund University
Imaging biomarker for prostate cancer
Evaluation of the Bone Scan Index
In collaboration with Lars Edenbrandt, Clinical Physiology, Malmö
![Page 10: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/10.jpg)
Astronomy & Theoretical Physics Lund University
This project deals with later stages of prostate cancer, specifically when bone metastases occur.
Aim: Characterize the BSI imaging biomarker
- prognosis- treatment response
Other common biomarkers or “scores” may not be optimal(e.g. PSA, Gleason score)
![Page 11: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/11.jpg)
Astronomy & Theoretical Physics Lund University
BSI = Bone Scan Index
BSI is a method of expressing the tumor burden in the bone as a percentage of the total skeletal mass.
bsi < 1 1 < bsi < 5 bsi > 5
![Page 12: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/12.jpg)
Astronomy & Theoretical Physics Lund University
How to calculate BSI?
![Page 13: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/13.jpg)
Astronomy & Theoretical Physics Lund University
Imageregistration
Hotspot detection
Hotspot features
Hotspotclassification
BSIcalculation
Imageanalysis
Machinelearning
Whole-body bone scan
2D → 3Dmodel
![Page 14: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/14.jpg)
Astronomy & Theoretical Physics Lund University
Atlas
• Based on ~10 images• Manual delineation• 12 anatomical regions
The atlas is registred tothe new image using the
Morphon method
Imageregistration
![Page 15: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/15.jpg)
Astronomy & Theoretical Physics Lund University
• Find average intensity in healthy bone tissue
• Normalize using the above average intensity
• Bandpass filtering
• Thresholding
Hotspot detection
• Geometry features
• Localization features
• Intensity distribution features
• Other global features capturing the density of hotspots. Both regional and global.
Hotspot features
Machinelearning
![Page 16: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/16.jpg)
Astronomy & Theoretical Physics Lund University
Now we can!
Exini Diagnostics
![Page 17: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/17.jpg)
Astronomy & Theoretical Physics Lund University
Data overview
Event or censoredPSA
Bone scan 1 Bone scan 2
PSA
Possible primary therapy
Time
![Page 18: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/18.jpg)
Astronomy & Theoretical Physics Lund University
COX proportional hazard model of survival data – very common method.
Relative risk (hazard ratio) becomes simple. For examplecomparing a unit change of one covariate:
How to model survival?
can be estimated using “maximum partial likelihood”
![Page 19: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/19.jpg)
Astronomy & Theoretical Physics Lund University
Some results
![Page 20: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/20.jpg)
Astronomy & Theoretical Physics Lund University
COX -modelanalysis
High risk patientsafter primary therapy
Serial analysis, i.echange of BSI
Regional analysis ofBSI measurements
![Page 21: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/21.jpg)
Astronomy & Theoretical Physics Lund University
Risk evaluation before heart transplantation
In collaboration with Johan Nilsson. hjärtkirurgi, LU
![Page 22: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/22.jpg)
Astronomy & Theoretical Physics Lund University
Overall survival for ~ 56 000transplantations
![Page 23: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/23.jpg)
Astronomy & Theoretical Physics Lund University
Recipient-Donor matching problem
Age (years)
Female gender
Height (cm)
Weight (kg)
Ischemic cardiomyopathy
Non-ischemic cardiomyopathy
Insulin-treated diabetes
Hypertension
Antiarrhythmic
Amiodarone
Previous blood transfusion
Previously transplanted
Previous cardiac surgery
ECMO
Blood group (A,B,AB,O)
Creatinine (µmol/l)
Serum bilirubin (mg/dl)
….
Demographic data
Age (year)
Female gender
Weight (kg)
Duration of ischemia (min)
CODD: Head trauma
CODD: Cerebrovascular event
Blood group (A,B,AB,O)
Recipient Donor
Optimalmatch?
Many One
In total about 140 available “features”
![Page 24: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/24.jpg)
Astronomy & Theoretical Physics Lund University
Today
Possible recipients
1. Compatible blood group match2. Recipient donor weight match ± 20%
Prioritize according to
1. Identical blood group2. If young donor, select young recipient (< 35 years) or donor age - recipient age < 15 years3. If PVR > 3.0 then 0-15% larger weight for the donor
Two or more recipients havethe same priority then
random selection
Aim: Better selection → improved survival
![Page 25: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/25.jpg)
Astronomy & Theoretical Physics Lund University
Beyond the COX model
Discrete hazard rate
Survival function
For discrete data
As usual “maximize the likelihood function”
![Page 26: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/26.jpg)
Astronomy & Theoretical Physics Lund University
PLANN = Partial logistic regression with ANN
The output from the neural network will provide smoothed estimatesof the discrete hazard rates.
A more flexible modeling!
Model these by neural networks
Also add: regularization, ensemble approaches, multiple random imputations etc.
![Page 27: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/27.jpg)
Astronomy & Theoretical Physics Lund University
● 56 625 patients that have undergone a heart transplantation
● Mean age was 51 and 21% women.
● Mean follow-up duration of 5.2 years
● Overall 30-day mortality was 9% (n=5010)
● One-year mortality was 18% (n=9380)
● A total of 21 502 patients (38%) died during follow-up.
● Main cause of death was
- late graft failure (3215)- major adverse cardiovascular events (2993)- infections (2656)
Study Population – ISHLT database
Also: Scandiatransplant, ~1300 patients, external validation
![Page 28: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/28.jpg)
Astronomy & Theoretical Physics Lund University
![Page 29: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/29.jpg)
Astronomy & Theoretical Physics Lund University
We now have a model!
![Page 30: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/30.jpg)
Astronomy & Theoretical Physics Lund University
![Page 31: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/31.jpg)
Astronomy & Theoretical Physics Lund University
Can we learn something new?
● The model gives you a predicted survival curve for a donor-recipient pair.
● We can measure the “performance” for anygiven pair (both real and virtual).
● The area under S(t) is our measureof performance.
![Page 32: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/32.jpg)
Astronomy & Theoretical Physics Lund University
Virtual recipient-donor matching
Recipient
Visualize the important combinationsusing a regression tree
Vr1
Vr2
Vr6
Vr5
Vr4
Vr3
Donor
Vd1
Vd2
Vd6
Vd5
Vd4
Vd3
Match allcombinations
![Page 33: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/33.jpg)
Astronomy & Theoretical Physics Lund University
![Page 34: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/34.jpg)
Astronomy & Theoretical Physics Lund University
C-index modeling
In-house development (Patrik E & Jonas K)
![Page 35: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/35.jpg)
Astronomy & Theoretical Physics Lund University
C-index (concordance index) is a performance measure for survival modeling (with censored data)
predicted survival time for patient i.
All pairs of patients (i,j) such that:● Both i and j have events and ti < tj● Patient i have and event and ti is smaller that patient j's censor time
You do not have to predict the survival time. A prognostic index that can orderthe patients correctly is sufficient.
![Page 36: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/36.jpg)
Astronomy & Theoretical Physics Lund University
The c-index is rank based measure = problems whenoptimizing the prediction model.
Our approach:
- Use neural networks to compute a prognostic index.
- Maximize the C-index with respect to model parameters
- We use genetic algorithms for the optimization.
- Use en ensemble of networks rather than just a single one.
![Page 37: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/37.jpg)
Astronomy & Theoretical Physics Lund University
Clinical application
● ~ 4000 female patients with breast cancer, that have had removal ofprimary tumor.
● Recurrence for about 21% (after 5 years)
● Median age ~ 60 years.
Example of covariates: Age, tumor size, number of positve lymph nodes, HER2-status, histological grade.
Aim: Construct a prognostic index for recurrence
![Page 38: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/38.jpg)
Astronomy & Theoretical Physics Lund University
Split into high risk and low risk group based on the index
![Page 39: Bild 1home.thep.lu.se/~anders/ATP_slides/140205-Mattias.pdfTitle: Bild 1 Author: mattias Created Date: 2/5/2014 4:28:48 PM](https://reader033.vdocuments.site/reader033/viewer/2022051908/5ffca97591ac0e2e6b2e3d24/html5/thumbnails/39.jpg)
Astronomy & Theoretical Physics Lund University
Extension: Predict actual survival times.
More extension: Use more information from the censored cases
Machine learning model
Correct for censored cases