Machine Learning /////////// Introduction May 2018 / Katja Gla - PowerPoint PPT Presentation

Machine Learning /////////// Introduction May 2018 / Katja Glaß

Agenda Overview Neural Networks CTCAE Grading as Example Use Cases Summary Machine Learning • May 2018 • Katja Glaß Page 2

Overview Machine learning is a field of computer science that gives computers the ability to learn without being explicitly programmed. Machine Learning • May 2018 • Katja Glaß Page 3

Overview Different approaches for different purposes Big data / high complexity Deep Neural Networks Recurrent Neural Networks Reinforcement Learning Machine Learning • May 2018 • Katja Glaß Page 4

Overview Different approaches for different purposes Artificial Neural Networks Classification Logic Regression Support Vector Machines Supervised Random Forest … Artificial Neural Networks Regresssion Linear Regression ML Decision Trees Bayesian Networks … Clustering Unsupervised Dimension Reduction Machine Learning • May 2018 • Katja Glaß Page 5

Overview Neural Networks Current Hype are Neural Networks & Deep Neural Networks Extremly powerful Image recognition Natural language processing Machine Learning • May 2018 • Katja Glaß Page 6

Neural Networks Rough understanding 1) Transform issues to numeric representation Number of inputs (neurons) Value of each input Example: Example Neurons Neuron Values Grey Image Each pixel one neuron Grey value Text Each word a neuron Frequency Lab analysis Each numeric variable Numeric value Each characteristic of 0 / 1 appearance character variable Machine Learning • May 2018 • Katja Glaß Page 7

Neural Networks Rough understanding 1) Transform issues to numeric representation Number of inputs (neurons) Value of each input Example: Example Neurons Neuron Values Elephant? (image) One Elephant 0 .. 1 (probability) Text classification (SDTM, Each characteristic one ADAM 0 .. 1 (probability) ADAM, TLF?) neuron SDTM 0 .. 1 (probability) TLF 0 .. 1 (probability) Grading value? 0..4 Classification -> one neuron Grade 0 0 .. 1 (prob) … per classification Machine Learning • May 2018 • Katja Glaß Page 8

Neural Networks Rough understanding 2) Structure Weight & Bias (optimized by algorithm) Number of neurons (5,3,4,1) Number of layers (3 with output layer) Machine Learning • May 2018 • Katja Glaß Page 9

Neural Networks Rough understanding 3) Mathematics Structure Activation Function (A) (output = f(input(s)) A A A A A A A A Machine Learning • May 2018 • Katja Glaß Page 10

Neural Networks Rough understanding 3) Mathematics Structure Activation Function (A) (output = f(input(s)) A Success-Function (B) A Cost / Utility A A A Accuracy A B Square Error … A A Machine Learning • May 2018 • Katja Glaß Page 11

Neural Networks Rough understanding 3) Mathematics Structure Activation Function (A) (output = f(input(s)) C A Success-Function (B) A Cost / Utility A A A Accuracy A B Square Error … A A Optimization function (C) – updates weights & bias Gradient Decent … Machine Learning • May 2018 • Katja Glaß Page 12

Neural Network Recommended Literature Free EBook “Neural Networks and Deep Learning” (http://neuralnetworksanddeeplearning.com/) Introduction & Statistical insights MNIST Example, large database with labeled handwritten digits Video Introduction “ Tensorflow and deep learning - without a PhD” (https://www.youtube.com/watch?v=vq2nnJ4g6N0&list=LL3uReggFn2MOSMZlG69vzEw) 2,5h video introduction on Tensorflow (free Google machine learning toolset) Top for first hands-on experiences, implementation complexity low Cool real world applications in a nutshell (~4 minutes each) https://www.youtube.com/watch?v=Bui3DWs02h4 https://www.youtube.com/watch?v=aKSILzbAqJs Machine Learning • May 2018 • Katja Glaß Page 13

Neural Network Machine Learning Crash Course Machine Learning Crash Course with TensorFlow APIs 40+ exercises 25 lessons 15 hours Lectures from Google researchers Real-world case studies Interactive visualizations of algorithms in action Machine Learning • May 2018 • Katja Glaß Page 14

CTCAE Grading as Example Anemia – simple test case First hands-on-experiences See how / why it is working Student (4 weeks) with Phython, Keras & TensorFlow experiences Two input parameters (Limit, Lab value) One output parameter (Grading 0 .. 3) Machine Learning • May 2018 • Katja Glaß Page 15

CTCAE Grading as Example Anemia – simple test case Input : LBSTNRLO (low), LBSTRESN (value) Ideal curve for LBSTNRLO = 12 XTTEST XTTESTCD XTSTRESN LBTEST LBTESTCD LBCAT LBSTRESU RANGE_LOW RANGE_HIG H 3 XTTEST XTTESTCD XTSTRESN LBTEST LBTESTCD LBCAT LBSTRESU lower end upper end Grading Value 2 Anemia BLANE 0 Hemoglobin HGB HEMATOLOGY g/dL GE LBSTNRLO 1 Anemia BLANE 1 Hemoglobin HGB HEMATOLOGY g/dL GE 10.0 LT LBSTNRLO 0 Anemia BLANE 2 Hemoglobin HGB HEMATOLOGY g/dL GE 8.0 LT 10.0 5 6 7 8 9 10 11 12 13 14 LBSTRESN Anemia BLANE 3 Hemoglobin HGB HEMATOLOGY g/dL LT 8.0 Machine Learning • May 2018 • Katja Glaß Page 16

CTCAE Grading as Example Network Layout Sequential 20/2/4 Sequential 3/4 Machine Learning • May 2018 • Katja Glaß Page 17

CTCAE Grading as Example Input Data Random Study Machine Learning • May 2018 • Katja Glaß Page 18

CTCAE Grading as Example Implementing a Model with Keras (TensorFlow) Machine Learning • May 2018 • Katja Glaß Page 19

CTCAE Grading as Example Results - Random generated Data Model Data Loop1 Loop2 Loop3 Range = 12 Accuracy Accuracy Accuracy Accuracy 20 / 2 Random 89.67% 93.33% 96.33% 92,28% Machine Learning • May 2018 • Katja Glaß Page 20

CTCAE Grading as Example Results - Random generated Data Model Data Loop1 Loop2 Loop3 Range = 12 Accuracy Accuracy Accuracy Accuracy 20 / 2 Random 89.67% 93.33% 96.33% 92,28% 3 Random 88.18% 96.37% 98.63 98,13% generated Data Machine Learning • May 2018 • Katja Glaß Page 21

CTCAE Grading as Example Results - Random generated Data Model Data Loop1 Loop2 Loop3 Range = 12 Accuracy Accuracy Accuracy Accuracy 20 / 2 Random 89.67% 93.33% 96.33% 92,28% 3 Random 88.18% 96.37% 98.63% 98,13% 3 Clinical 96.03% 98.91% 99.03% Machine Learning • May 2018 • Katja Glaß Page 22

CTCAE Grading as Example Results - Random generated Data Model Data Loop1 Loop2 Loop3 Range = 12 Accuracy Accuracy Accuracy Accuracy 20 / 2 Random 89.67% 93.33% 96.33% 92,28% 3 Random 88.18% 96.37% 98.63% 98,13% 3 Clinical 68.75% 96.03% 98.91% 99.03% Machine Learning • May 2018 • Katja Glaß Page 23

CTCAE Grading as Example Results - Random generated Data Model Data Loop1 Loop2 Loop3 Range = 12 Accuracy Accuracy Accuracy Accuracy 20 / 2 Random 89.67% 93.33% 96.33% 92,28% 3 Random 88.18% 96.37% 98.63% 98,13% 3 Clinical 68.75% 96.03% 98.91% 99.03% Machine Learning • May 2018 • Katja Glaß Page 24

Use Cases Where can we apply Machine Learning in our area? Machine Learning • May 2018 • Katja Glaß Page 25

Use Cases CSS Working Group „ Machine Learning“ Educate for Future! Find use cases Drug discovery Drug candidate selection Clinical system optimization Medical image recognition Medical diagnoses Optimum site selection / recruitment Data anomaly detection Personalized medicine Machine Learning • May 2018 • Katja Glaß Page 26

Use Cases More Use Cases Data mapping Coding Virtual monitoring / studies (medical devices) RWE evaluations (tons of data) Chat bots for questions / tickets Document search Machine Learning • May 2018 • Katja Glaß Page 27

Use Cases CTCAE Grading Apply ML on available formula for learning (1) Investigate depending parameters for doctor’s required grading (2) Check doctor’s grading (3) Machine Learning • May 2018 • Katja Glaß Page 28

Use Cases CTCAE Grading Apply ML on available formula for learning (1) Investigate depending parameters for doctor’s required grading (2) Check doctor’s grading (3) Positive Negative Challenge doctor’s decision? Learning opportunity with hands-on experiences (1) Lower machine-power / Laptop sufficient (1) (no clould required) Enough data available Machine Learning • May 2018 • Katja Glaß Page 29

Use Cases Optimum site selection / recruitment Build up site / recruitment data pool ML to learn quality / expected recruitment on parameters Machine Learning • May 2018 • Katja Glaß Page 30

Use Cases Optimum site selection / recruitment Build up site / recruitment data pool ML to learn quality / expected recruitment on parameters Challenges Get information of possible parameters/features General (no. of staff, fluctuation rate, urban catchment, …) Study specific (TAS, experts, experiences, past recruitment- time rate, …) Subject specific (dropout rate, screening failure rate, … ) Evaluation size, Cloud need Likely need “machine learning” optimized machine Machine Learning • May 2018 • Katja Glaß Page 31