Department of Information Technology
V R Siddhartha Engineering College
Genomic Risk Assessment and Early Intervention for Rare Genetic
Disorders.
B.Tech in Information Technology
Presented by
N.Ram Tarun (218w1a1240)
S.Sashank Sai (218w1a1256)
K.Vaibhav (218w1a1225)
Under the guidance of
Dr G. Kalyani, Associate Professor
PROBLEM STATEMENT
Accurate genetic disease classification is crucial for timely diagnosis and effective
treatment. Misdiagnosis or delayed diagnosis can lead to poor patient outcomes and
significant healthcare costs. The complexity of genetic disorders and their interactions
with environmental factors hinders the development of robust classification models. A
reliable and accurate genetic disease identification system is essential to address this
challenge.
OBJECTIVES
1. Early Detection: Identify individuals at risk of developing a genetic disorder at an early stage.
2. Precision: Customize predictions based on individual genetic profiles.
3. Accuracy: Develop a highly accurate predictive model to minimize false positives and false
negatives.
4. Diagnostic Aid: Serve as a diagnostic aid for healthcare professionals by providing genetic
basis insights.
OUTCOMES
• Early detection and intervention for rare genetic disorders
• Empowering patients and families thorough access to genetic risk assessments .
• Contribute to broader research and clinical advancements, benefiting the field of
genetics and healthcare.
• Tailored interventions and therapies to individual genetic profiles, potentially improving
treatment efficacy and minimizing adverse effects.
DATASET DESCRIPTION
GENETIC VARIANT CLASSIFICATIONS (KAGGLE)
• Clinvar is a public archive of reports about the relationships between human genetic variations and phenotypes, with
supporting evidence
•The dataset focuses on categorizing variant alleles and their interpretation from a clinical standpoint
•65189 ROWS 46 COLUMNS
GENOMES AND GENETIC (BIOPORTAL)
• The dataset contains two files: train.Csv with 22083 rows and 45 columns, and test.Csv with 9465 rows and 43 columns
• The dataset attributes include patient id, genetic disorder, disorder subclass, patient age, and various medical test results
REQUIREMENTS
• DATASETS TO BUILD THE MODEL
GENETIC VARIANT CLASSIFICATION
GENOMES AND GENETICS
• SOFTWARE REQUIREMENTS
GOOGLE COLABORATORY / JUPYTER NOTEBOOK
TORCH LIBRARY (PYTORCH)
ALGORITHMS USED
1. SUPPORT VECTOR MACHINE (SVM)
2. RANDOM FOREST CLASSIFIER (RFC)
3. ARTIFICIAL NEURAL NETWORKS (ANN)
ARCHITECTURE DIAGRAM
RESULTS
1. Model Evaluation:
- Support Vector Machine (SVM) : - Accuracy: 91%
- Random Forest : - Accuracy: 93%
- Artificial Neural Network (ANN) : - Accuracy: 90%
2. Conclusion:
- The implemented models, including SVM, Random Forest, and ANN, demonstrated strong predictive capabilities
for Genetic Disorders.
- The Random Forest Classifier outperformed the other models, achieving the highest accuracy of 93%.
- These results indicate the potential of machine learning in accurately predicting genetic disorders based on
patient data.
- The models can be valuable tools for early detection and intervention, contributing to improved patient outcomes.
The following table shows the accuracy of each model on the test set:
ALGORITHM SVM ANN RANDOM
FOREST
ACCURACY 91% 90% 93%
RESULTS
The Random Forest Classifier, with a
random state of 42, was evaluated using
precision, recall, and f1-score metrics for
each class in the genetic disease dataset.
The classifier showed an impressive overall
accuracy of 91%, with high precision and
recall values for individual classes,
demonstrating its efficacy across diverse
classes.
CLASSIFICATION CONFUSION
REPORT MATRIX
IMPACT TO THE SOCIETY
• The project can lead to informed medical decisions, improved healthcare outcomes, and a
reduction in the prevalence of severe genetic disorders within affected families. It aligns with
the emerging field of personalized medicine, which tailors treatments to an individual's genetic
profile.
• Community and family well-being
• Improved healthcare accessibility
• Ethical considerations and informed decision-making
FIELD WORK
Dr.KOTI KALYAN, Pathologist
Gayatri Super Speciality Hospital
REFERENCES
1. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6287949/
2. https://pubmed.ncbi.nlm.nih.gov/36672812/#:~:text=This%20study%
20focuses%20on%20the,feature%20set%20for%20model%20training
3. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9173933/
4. https://humgenomics.biomedcentral.com/articles/10.1186/s40246-019
-0215-5
5. https://genomemedicine.biomedcentral.com/articles/10.1186/gm359
