Data-driven Automatic Treatment Regimen Development and - - PowerPoint PPT Presentation

▶
data driven automatic treatment regimen development and
SMART_READER_LITE
LIVE PREVIEW

Data-driven Automatic Treatment Regimen Development and - - PowerPoint PPT Presentation

Aug 22, 2023 370 likes •757 views

Data-driven Automatic Treatment Regimen Development and Recommendation Date: 2016/09/20 Author: Leilei Sun, Chuanren Liu, Chonghui Gou, Hui Xiong, Yatming Xie Source: ACM KDD 16 Advisor: Jia-ling Koh Speaker : Yi-hui Lee 1 Outline

slide-1
SLIDE 1

Date: 2016/09/20 Author: Leilei Sun, Chuanren Liu, Chonghui Gou, Hui Xiong, Yatming Xie Source: ACM KDD’ 16 Advisor: Jia-ling Koh Speaker : Yi-hui Lee

1

Data-driven Automatic Treatment Regimen Development and Recommendation

slide-2
SLIDE 2

Outline

  • Introduction
  • Approach
  • Experiment
  • Conclusion

2

slide-3
SLIDE 3

Introduction

  • Motivation:

The typical treatment regimens are usually used as prototypes when a clinical doctor designs the personalized treatment plan for a new patient.

  • Goal:

Typical treatment regimens automatic identification and evaluation.

3

slide-4
SLIDE 4
  • Treatment regimen recommendation framework:

4

Introduction(cont.)

step 1 step 2 step 3 step 4

slide-5
SLIDE 5
  • Data:

EMRs: Electronic Medical Records

  • Challenges:

Complex EMRs data: effective method to measure the similarity between treatment records Large volume of patients’ records: clustering algorithm->MRDPC

5

Introduction(cont.)

slide-6
SLIDE 6

Outline

  • Introduction
  • Approach
  • Experiment
  • Conclusion

6

slide-7
SLIDE 7

Approach

7

  • Treatment regimen recommendation framework:

step 1 step 2 step 3 step 4

7

slide-8
SLIDE 8

Approach(cont.)

8

  • Data:

EMRs: Contain five categories of information of patients

  • Demographic information
  • Diagnostic information
  • Laboratory indicators
  • Doctor orders
  • Outcomes
slide-9
SLIDE 9

Approach(cont.)

9

  • Demographic information: Age, gender, address,

race and ethnicity, education, and other information

  • f a patient.
slide-10
SLIDE 10

Approach(cont.)

10

  • Diagnostic information(given by doctors): It

consists of disease names and severity of the diseases.

slide-11
SLIDE 11

Approach(cont.)

11

  • Laboratory indicators: Mainly used for judging the

severity of disease or evaluating a patient’s final

  • utcome, which can be implied by diagnostic

information or the outcome, so which are not included in the following model.

slide-12
SLIDE 12

Approach(cont.)

12

  • Treatments: Series of doctor orders
  • Doctor orders: Medical prescription

Drug Name

slide-13
SLIDE 13

Approach(cont.)

13

  • Treatments: Series of doctor orders
  • Doctor orders: Medical prescription

Delivery Route

IV: Intravenous injection IM: Intramuscular Per os, PO: Oral

slide-14
SLIDE 14

Approach(cont.)

14

  • Treatments: Series of doctor orders
  • Doctor orders: Medical prescription

Dosage each time

slide-15
SLIDE 15

Approach(cont.)

15

  • Treatments: Series of doctor orders
  • Doctor orders: Medical prescription

Frequency frequency per day

slide-16
SLIDE 16

Approach(cont.)

16

  • Treatments: Series of doctor orders
  • Doctor orders: Medical prescription

Repeating times during the q-th period

slide-17
SLIDE 17

Approach(cont.)

17

  • Outcome: Evaluated and presented by doctors

when a patient leaves hospital

  • Cured
  • Improved
  • Ineffective
  • Dead
slide-18
SLIDE 18

Approach(cont.)

18

  • Similarity Measure for Treatment
  • s̄q(i,j) is the similarity of OSPiq and OSPjq, which is

the similarity of two treatments in q-th period.

w = (0.4 0.2 0.2 0.2)

fiqg is the repetition times of g-th order in q-th period of treatment i {aijqgh} is the allocation matrix for the computing of |OSPiq ∩ OSPjq |

slide-19
SLIDE 19

Approach(cont.)

19

  • Similarity Measure for Treatment

A B C D E

S(A, D)=(1x[0+120/160])/2 S(A, E)=(0x[1+24/160])/2 S(B, D)=(0x[0+18/120])/2 …

DN DE

DD= ODose × OFreq

DN DE

DD= ODose × OFreq

slide-20
SLIDE 20

Approach(cont.)

20

  • Similarity Measure for Treatment
  • Properties:
  • 1) The value of s ̄
q(i,j) is from 0 to 1
  • 2) Symmetry. For any i and j, s ̄
q(i,j) = s ̄ q(j,i);
  • 3)Self-similarity. s ̄
q(i,j)=1,if and only if OSPiq =OSPjq
slide-21
SLIDE 21

Approach(cont.)

21

  • Clustering Treatments: Map Reduce enhanced

Density Peaks based Clustering (MRDPC)

  • DPC: Density Peaks based Clustering
  • Derives from exemplar-based clustering algorithm
  • Discover clusters with complex shapes
slide-22
SLIDE 22

Approach(cont.)

22

  • DPC

Compute two indicators for each object 1) Local density ρ meaning of ρi is to count the number of objects in object i’s sc- neighborhood 2) Minimum distance (or maximum similarity) between the

  • bject and any other object with higher local density γ

Objects with larger ρ and lower γ values are viewed as exemplars.

χ(x)=1 if x>0 and χ(x)=0 otherwise

slide-23
SLIDE 23
  • MRDPC:

The total N patients are first randomly divided into m parts

Approach(cont.)

23

p1 p2 p3 p4 pN

… Parts1 Parts2 Partsm …

p1 pN p3 p9

… … …

DPC

get k potential exemplars

Parts1 Parts2 Partsm …

p1 p9

… … …

slide-24
SLIDE 24
  • Extracting Typical Treatment Regimens:

A typical treatment regimen includes the names

  • f medicines used in a specified period, the

dosages, the delivery routes, and lasting how many days. Extract a semantic description of each treatment cluster by its dense core. Dense core: Constructed by k-nearest neighbors

  • f its exemplar.

Approach(cont.)

24

slide-25
SLIDE 25
  • Extracting Typical Treatment Regimens:

Approach(cont.)

25

A treatment regimen Specified period

λ(Drug, OSP ) = 1 if Drug is used in OSP (Drug ∈ OSP ), λ(Drug, OSP ) = 0 otherwise DA is a triple consists of delivery route, dosage and lasting days of a order

slide-26
SLIDE 26
  • Evaluating Typical Treatment Regimen:

Patient cohort: The patients in a same leaf node is defined as a patient cohort

  • Decision tree model
  • Divide patients into different groups according to

demographic information, diagnostic information and outcomes.

Approach(cont.)

26

slide-27
SLIDE 27

Approach(cont.)

27

  • Treatment regimen recommendation framework:

step 1 step 2 step 3 step 4

27

slide-28
SLIDE 28

Outline

  • Introduction
  • Approach
  • Experiment
  • Conclusion

28

slide-29
SLIDE 29

Experiment

29

  • Experimental Data: EMRs data used in this paper

are collected from Hos- pital Information Systems (HIS) of 14 Grade Three Class A (G3CA) hospitals Focus on the patients with cerebral infarction disease, which is one of the most common diseases in China today Extract the typical treatment regimens of cerebral infarction from doctor orders of 28, 659 patients. The total number of doctor orders is 1, 007, 057

slide-30
SLIDE 30

Experiment(cont.)

  • Extracting Typical Treatment Regimens:
  • Select 138 medicines that are most relevant to

cerebral infarction

  • 363,674 doctor orders containing the selected

medicines, nearly 13 doctor orders per patient.

30

slide-31
SLIDE 31

Experiment(cont.)

  • Extracting Typical Treatment Regimens:

31 Four typical treatment regimens extracted from EMRs

slide-32
SLIDE 32

Experiment(cont.)

  • Extracting Typical Treatment Regimens:

32 An example of an extracted treatment regimen

slide-33
SLIDE 33

Experiment(cont.)

  • Treatment Regimen Recommendation:

33 Recommend treatment regimens for two patient cohorts

slide-34
SLIDE 34

Experiment(cont.)

  • Treatment Regimen Recommendation:

34

slide-35
SLIDE 35

Experiment(cont.)

  • Overall Treatment Evaluation:

35 Our method can help improve effective rate and cure rate

slide-36
SLIDE 36

Outline

  • Introduction
  • Approach
  • Experiment
  • Conclusion

36

slide-37
SLIDE 37

Conclusion

37

  • Investigated how to identify the typical treatment regimens from

large-scale treatment records and how to find the most effective treatment regimens for patients. Developed an efficient semantic clustering algorithm, based on a new method to measure the similarities between treatment records. Applied on large-scale treatment records, we were able to extract the treatment clusters as the typical treatment regimens with semantically meaningful descriptions. Designed a unified framework to evaluate the effectiveness of the identified treatment regimens

  • This work may be the first step towards the automatic development
  • f treatment regimens and treatment recommendations.