A Case Study in Machine Learning Harikrishna Narasimhan Department - - PowerPoint PPT Presentation

▶
a case study in machine learning
SMART_READER_LITE
LIVE PREVIEW

A Case Study in Machine Learning Harikrishna Narasimhan Department - - PowerPoint PPT Presentation

Sep 23, 2023 231 likes •673 views

Predicting Anticancer Drug Response A Case Study in Machine Learning Harikrishna Narasimhan Department of Computer Science and Automation I Indian Institute of Science, Bangalore Joint work with Shivani Agarwal, IISc and Mitra Biotech An

slide-1
SLIDE 1

Predicting Anticancer Drug Response – A Case Study in Machine Learning

Joint work with Shivani Agarwal, IISc and Mitra Biotech

Harikrishna Narasimhan

Department of Computer Science and Automation I Indian Institute of Science, Bangalore

slide-2
SLIDE 2

Predictive Model

An Example Machine Learning System

slide-3
SLIDE 3

Predictive Model

malignan gnant t

  • r

benign ?

Patient X-ray with Lung Tumor

An Example Machine Learning System

slide-4
SLIDE 4

malignan gnant benign gn

…

benign gn

An Example Machine Learning System

slide-5
SLIDE 5

malignan gnant benign gn

…

Machine Learning Algorithm

Predictive Model

benign gn

An Example Machine Learning System

slide-6
SLIDE 6

malignan gnant benign gn

…

Machine Learning Algorithm

Predictive Model

benign gn

An Example Machine Learning System

slide-7
SLIDE 7

Predicting Anticancer Drug Response A Case Study

Majumder, B., Baraneedharan, U., Thiyagarajan, S., Radhakrishnan, P., Narasimhan, H., Dhandapani, M., Brijwani, N., Pinto, D.D., Prasath, A., Shanthappa, B.U., Thayakumar, A., Surendran, R., Babu, G., Shenoy, A.M., Kuriakose, M.A., Bergthold, G., Horowitz, P., Loda, M., Beroukhim, R., Agarwal, S., Sengupta, S., Sundaram, M. and Majumder, P.K. Predicting clinical response to anticancer drugs using an ex vivo platform that captures tumor heterogeneity. Nature Communications. To appear.

slide-8
SLIDE 8

Pre-dose Post-dose CR NR PR

Pre-dose Post-dose Courtesy: Mitra Biotech

slide-9
SLIDE 9

Pre-dose Post-dose CR NR PR

Complete Responder Pre-dose Post-dose Courtesy: Mitra Biotech

slide-10
SLIDE 10

Pre-dose Post-dose CR NR PR

Complete Responder Partial Responder Pre-dose Post-dose Courtesy: Mitra Biotech

slide-11
SLIDE 11

Pre-dose Post-dose CR NR PR

Complete Responder Partial Responder Non- Responder Pre-dose Post-dose Courtesy: Mitra Biotech

slide-12
SLIDE 12

Drug regimen 1 Drug regimen 2 Drug regimen 3 . . . . Drug regimen N

Predicting Anticancer Drug Response

slide-13
SLIDE 13

Drug regimen 1 Drug regimen 2 Drug regimen 3 . . . . Drug regimen N Complete Response (CR) Partial Response (PR) No Response (NR)

  • 100

100

Predicting Anticancer Drug Response

slide-14
SLIDE 14

Drug regimen 1 Drug regimen 2 Drug regimen 3 . . . . Drug regimen N

Predicting Anticancer Drug Response

an ex vivo platform for personalized cancer treatment

slide-15
SLIDE 15

Drug regimen 1 Drug regimen 2 Drug regimen 3 . . . . Drug regimen N

Tumor Ecosystem

Predicting Anticancer Drug Response

an ex vivo platform for personalized cancer treatment

slide-16
SLIDE 16

Drug regimen 1 Drug regimen 2 Drug regimen 3 . . . . Drug regimen N

Tumor Ecosystem

Drug regimen

Predicting Anticancer Drug Response

an ex vivo platform for personalized cancer treatment

slide-17
SLIDE 17

Drug regimen 1 Drug regimen 2 Drug regimen 3 . . . . Drug regimen N

Tumor Ecosystem

Drug regimen Parameters

Predicting Anticancer Drug Response

an ex vivo platform for personalized cancer treatment

slide-18
SLIDE 18

Drug regimen 1 Drug regimen 2 Drug regimen 3 . . . . Drug regimen N

Tumor Ecosystem

Drug regimen Parameters

Predict response

Predicting Anticancer Drug Response

an ex vivo platform for personalized cancer treatment

slide-19
SLIDE 19

Predictive Model

Predictive Model

respond

  • nder
  • r
  • r

non

  • n-resp

espond

  • nder?

Tumor Ecosystem Drug regimen Parameters

slide-20
SLIDE 20

Predictive Model

Predictive Model

respond

  • nder
  • r
  • r

non

  • n-resp

espond

  • nder?

Tumor Ecosystem Parameters:

Viability Histology Proliferation Apoptosis

Drug regimen

slide-21
SLIDE 21

Predictive Model

Predictive Model

respond

  • nder
  • r
  • r

non

  • n-resp

espond

  • nder?

Tumor Ecosystem Parameters:

Viability Histology Proliferation Apoptosis

Drug regimen

w1 x Viability + w2 x Histology + w3 x Proliferation + w4 x Apoptosis

slide-22
SLIDE 22

res esponder (R) res esponder (R) non-res esponder (NR) non-res esponder (NR)

Evaluating the predictive model

predicted actual

slide-23
SLIDE 23

res esponder (R) res esponder (R) non-res esponder (NR) non-res esponder (NR)

True Negative True Positive

Evaluating the predictive model

predicted actual

slide-24
SLIDE 24

res esponder (R) res esponder (R) non-res esponder (NR) non-res esponder (NR)

True Negative False Positive True Positive

Evaluating the predictive model

predicted actual

slide-25
SLIDE 25

res esponder (R) res esponder (R) non-res esponder (NR) non-res esponder (NR)

True Negative False Positive False Negative True Positive

Evaluating the predictive model

predicted actual

slide-26
SLIDE 26

res esponder (R) res esponder (R) non-res esponder (NR) non-res esponder (NR)

True Negative False Positive False Negative True Positive

Evaluating the predictive model

predicted actual

slide-27
SLIDE 27

res esponder (R) res esponder (R) non-res esponder (NR) non-res esponder (NR)

True Negative False Positive False Negative True Positive

Goal: Maximize true positives! Keep false positives in an acceptable range!

Evaluating the predictive model

predicted actual

slide-28
SLIDE 28

Evaluating the predictive model

slide-29
SLIDE 29

Evaluating the predictive model

slide-30
SLIDE 30

Evaluating the predictive model

slide-31
SLIDE 31

Evaluating the predictive model

0.25 ROC curve

slide-32
SLIDE 32

Evaluating the predictive model

0.25

good

  • dness

ness of

  • f

mo model

ROC curve

slide-33
SLIDE 33

SVMpAUC A New Machine Learning Method

  • Optimize performance in initial

portion of the curve

– Narasimhan H and Agarwal S, ICML 2013 – Narasimhan H and Agarwal S, KDD 2013

0.25

slide-34
SLIDE 34
  • 164 patients (109 training / 55 test)

– Head-and-neck cancer (HNSCC) – Colorectal cancer (CRC)

Experimental Results

slide-35
SLIDE 35
  • 164 patients (109 training / 55 test)

– Head-and-neck cancer (HNSCC) – Colorectal cancer (CRC)

  • Results: 100% true positive rate @ 25% false positive rate

Experimental Results

slide-36
SLIDE 36
  • 164 patients (109 training / 55 test)

– Head-and-neck cancer (HNSCC) – Colorectal cancer (CRC)

  • Results: 100% true positive rate @ 25% false positive rate

Experimental Results

slide-37
SLIDE 37
  • 164 patients (109 training / 55 test)

– Head-and-neck cancer (HNSCC) – Colorectal cancer (CRC)

  • Results: 100% true positive rate @ 25% false positive rate
  • Gives higher true positive rate than baseline SVOR machine

learning method

Experimental Results

slide-38
SLIDE 38

Experimental Results

slide-39
SLIDE 39

Experimental Results

slide-40
SLIDE 40

Colorectal Cancer (CRC)

Experimental Results

slide-41
SLIDE 41

Colorectal Cancer (CRC) Head and Neck Squamous Cell Carcinoma (HNSCC)

Experimental Results

slide-42
SLIDE 42

Acknowledgements

  • Pradip K. Majumder, Mitra Biotech, Bangalore
  • Biswanath Majumder, Mitra Biotech, Bangalore
  • Padhma Radhakrishnan, Mitra Biotech, Bangalore
  • Shiladitya Sengupta, Brigham and Women’s Hospital, Harvard

Medical School, Boston,USA

  • Mallikarjun Sundaram, Mitra Biotech, Bangalore