Enhancing Outcomes: Predictive Modeling for Breast Cancer Survival

In this presentation, Akhil explores innovative approaches to improving breast cancer survival rates through predictive modeling. By analyzing comprehensive patient data and leveraging advanced statistical techniques, Akhil will showcase how predictive models can forecast survival probabilities and identify key factors influencing outcomes. The presentation will cover the development and validation of these models, their implications for personalized treatment plans, and how they can aid healthcare professionals in making informed decisions. Attendees will gain insights into cutting-edge methods for enhancing patient care and advancing research in breast cancer survival. for more information visit: https://bostoninstituteofanalytics.org/data-science-and-artificial-intelligence/

1. Personalized
Treatment Strategies
for Breast Cancer
Using Survival
Analysis

Presented By: Akhil Akshay

B I A :: Revue OF
ANALYTICS

2. Agenda

Exploring Personalized Treatment Strategies for Breast Cancer Using Survival Analysis

Introduction to
Personalized Treatment
and Survival Analysis

An overview of how personalized
"treatment approaches enhance survival
‘analysis for breast cancer patients.

Model Evaluation and
Comparison
Methods for evaluating model

effectiveness and comparing diferent
approaches to find the best fit.

Data Preprocessing.
Techniques

Discuss techniques orcicaingand
Organ dato prepare for analy
ensuing accuracy

Advanced Statistical
Techniques

Insight into advanced statistical methods
that enhance the analysis and
inter

Feature Scaling and
Validation

Explore methods for feature scaling and
validation to improve model performance
land reliability

Key Findings and Future
Research Directions

‘Summarize the main findings and discuss
potential future research avenues in
personalized treatment

Model Selection and
| Training
Cover various models suitable for survival

analysis and the training processes for
‘optimal results

Boston ©
INSTITUTE OF
ANALYTICS

3. MISSING VALUE ANALYSIS

Missing Values Per Feature

Visualization of missing values across different features in the dataset.

30
25
22
18
10
a e we [
Neoplasm HER? status Hormone Inferred Tumor Size
Histologic Grade measured by Therapy Menopausal State
SNP6

BOSTO!
INSTITU
ANALYT

4. DATA PREPROCESSING

Data Preprocessing: Handling Missing Values and ean Features

Essential Steps for Effective Data Preparation in Survival Analysis

à

®
on

Checking and Visualizing Missing Handling Missing Values in
Values Numerical Data

Conduct an initial assessmentto Use median imputation to replace
identify missing data utilizing bar plots
to visualize the extent of missingness osuringminimalimpacton data

across features. distribution.

missing values in numerical columns,

RN;

Handling Missing Values in
Categorical Data

Replace missing values in categorical

Feature Scaling Importance

\dardScalerto normalize

tures, promoting.
columns with the mode, ensuring the y and improving model

most common category is retained. performance,

5. MODEL EVALUATION

Model Selection: Classification Algorithms

BOSTON ©
INSTITUTE OF
| ANALYTICS

Evaluating Algorithms for Personalized Breast Cancer Treatment

Logistic Regression ‘Support Vector Machine Decision Tree Random Forest Gradient Boosting
Astatistical method for predicting (SVM) Amodelthatuses atree-ike graph Combines multiple decision trees An ensemble technique that
binary outcomes, useful for initial Effectveforhigh dimensional ofdecisions,easytointerpretíor _toimprove accuracy and reduce — builds models sequentially to
insights data, SVM can separate classes treatment options. overfiting minimize errors effectively.

with ahy

6. EVALUATION METRICS

Model Evaluation Metrics

A Comparative Analysis of Various Machine Leaming Models for Breast Cancer Treatment

Logistic Regression

Decision Tree

Random Forest

Gradient Boosting

078

085

087

091

081

093

093

078

om

079

086

085

078

079

081

7. SURVIVAL ANALY...
Kaplan-Meier
Survival Analysis

Understanding Survival Rates for
Tailored Breast Cancer Treatments

8. SURVIVAL ANALYSIS
Kaplan-Meier Survival Curve

Visualization of survival probabilities over time with confidence intervals.

06 Months. 612 Months 1224 Months 24-36 Months 36-48 Months 8-60 Month

9. [SURVIVAL ANALYSIS INSIGHTS

Key Findings

Insights from Survival Analysis and Predictive Modeling in Breast Cancer Treatment

Survival Analysis insights
Kaplan-Meier curves highlight significant factors impacting patient
survival rates, aiding in treatment planning

Machine Learning Success

‘Advanced machine learning models demonstrate efetiveness in
predicting 10-year mortality risk for breast cancer patients.

Cox Model Analysis
‘Cox regression identifies key features that increase mortality risk,
providing valuable insights for clinicians.

Top Performing Models
Gradient Boosting and Random Forest models outperform others,
offering high accuracy in mortality sk predictions.

10. Genomic Data Integration

Enhancing predictions with genetic
‘markers like BRCA mutations can lead to
tailored therapies.

‘Advanced Machine Learning

Exploring deep learning models can
improve prediction accuracy for patient
‘outcomes.

Extended Longitudinal studies
Uniting larger datasets enhances
survival rend analysis over extended
periods.

Patient-Centric Models
Incorporating patientreported
‘outcomes fosters holistic treatment

strategies and personalized care

RESEARCH DIRECTIONS

Future Research Directions

Exploring Innovative Strategies for Breast Cancer Treatment

BIA

BOSTON ©
INSTITUTE OF
ANALYTICS

11. =

Survival Analysis Insights : Kaplan-Meier curves provided valuable insights into
patient survival rates over time, showing that certain factors, such as tumor size,
lymph node involvement, and molecular subtypes (ER/PR status), had a
significant impact on long-term survival.

BOSTON ©
INSTITUTE OF
ANALYTICS

12. =

Questions ......272222?

B | A :: Rene OF
ANALYTICS

13. Thank You!!!

B | AR: Name OF
ANALYTICS