SRM INSTITUTE OF SCIENCE AND TECHNOLOGY

RAMAPURAM
FACULTY OF ENGINEERING AND TECHNOLOGY
DEPARTMENT OF INFORMATION TECHNOLOGY
18CSP107L-Minor Project
8.11.2024
Medicine recommendation system using machine learning
BATCH MEMBERS
Shubham Kumar (RA2111008020128)
Sumit Kumar (RA2111008020129)
R Adhitya (RA2111008020163)

PROJECT GUIDE
Mrs. Rama S, AP/IT
 ABSTRACT

• This project tackles the challenge of suggesting personalized medications for patients based on
their medical history and symptoms . It leverages machine learning algorithms Support Vector
Classifier (SVC). With SVM having the highest accuracy of 1 to analyze extensive datasets that
include clinical trial data, and drug information By integrating these algorithms, the system
generates tailored medication suggestions aimed at improving patient outcomes and reducing
prescription errors .
• Ultimately, the project aspires to enhance treatment effectiveness and safety, fostering a more
personalized approach to healthcare and contributing to a more efficient healthcare system.
 PROBLEM IDENTIFICATION

PROBLEM STATEMENT:
• Healthcare struggles with providing personalized medication recommendations, often resulting in
ineffective treatments and adverse drug reactions due to diverse patient needs.

SOLUTION:
• A machine learning-based Medicine Recommendation System using Support Vector Classifier
(SVC) analyzes patient data and symptoms to suggest personalized medications. The system
improves over time, ensuring precise, real-time drug recommendations to enhance treatment
accuracy and reduce errors, ultimately supporting better patient outcomes.
 MOTIVATION OF THE PROJECT

• Enhanced Access to Healthcare: Aims to improve healthcare access by providing users, especially
those in remote or underserved areas, with initial medical advice and recommendations based on
their symptoms.

• Personalized Healthcare Recommendations: Utilizes machine learning to deliver customized

suggestions for medications, exercise regimens, dietary guidelines, and safety measures, supporting
individualized patient care.

• Reducing Healthcare Burden: Helps minimize unnecessary doctor visits by empowering users with
preliminary health insights, which can alleviate the strain on healthcare resources.
 EXISTING SYSTEM

• Uses predefined rules and expert knowledge to offer recommendations but struggle with complex
medical cases and adapting to new information.
• Existing systems often lack high-quality and diverse datasets, affecting accuracy and the ability to
represent complex health conditions.
• Many existing systems have complex interfaces that aren’t user-friendly, limiting accessibility for
non-specialists or general users seeking preliminary health advice.
 DISADVANTAGES

• Limited Personalization. Current systems are not yet capable of personalizing suggestions to
unique patient characteristics that could end up in generic suggestion.
• Manual Intervention: A high amount of manual intervention leads to the possibility of having
higher possibilities of errors and more time required to complete the recommendation.
• Scalability Issues: Such systems simply cant cope with big datasets or adapt to the complexity of
modern healthcare needs.
• Static Rules: The rule-based approach does not include any new medical knowledge or updated
treatment guidelines.
 PROPOSED SYSTEM

• •A proposed machine learning-based medical recommendation system would involve

collecting and processing patient data, including medical history, symptoms, and treatment
outcomes, to train predictive models. These models would analyze new patient data to provide
personalized recommendations for diagnoses and treatment plans.
• It will continuously learn and improve from feedbacks coming from patient outcomes and
clinician input. The model should prioritize interpretability to help the healthcare providers
understand and trust the recommendations. Meanwhile, it should ensure high levels of data
privacy and security for the protection of the information of the patients.
 ADVANTAGES

• It provides many advantages, such as personalizing treatment plans according to the individual
patient data, better diagnostic accuracy because of the recognition of patterns in large datasets, and
learning continuously from new data for improved recommendations. It also hastens decision-
making, detects rare conditions, and minimizes human error.
• These systems also provide a tremendous amount of medical research that is integrated and
evidence-based, making recommendations up-to-date and comprehensive, leading to better patient
outcomes and efficient healthcare delivery.
 SYSTEM REQUIREMENTS

• High-performance servers with multi-core processors (e.g., Intel Xeon or AMD) for training and
deploying machine learning models.
• Graphics Processing Units (GPUs) such as NVIDIA or AMD.
• RAM: Minimum of 12GB RAM for handling large datasets and complex computations.
• Operating System: Linux preferred for its stability and support for machine learning frameworks
(e.g., Ubuntu, CentOS).
 MODULES

•Data Ingestion
•Preprocessing
•Evaluation
•Visualization
•Deployment module
Data Collection:
•The dataset is gathered from sources (Kaggle) .
•The dataset contains information over 4000 + symptoms , drug information and clinical data.
Preprocessing:
•Cleaning – The system will remove duplicates, irrelevant data from the dataset.
•Feature transformation : Text data such as symptoms and descriptions are transformed into numerical
representations using techniques like TF-IDF or embeddings.
Data splitting:
•The dataset is divided into training and test sets (80% for training and 20 % for testing)
 SAMPLE
SCREENSHOTS
 SAMPLE
SCREENSHOTS
 SAMPLE
SCREENSHOTS
 CONCLUSIO
N
It proposed an anti-biotic medication suggestion system, through machine learning support vector
classification, such that by improving access to healthcare along with medication advise, and
lifestyle and preventative care mainly in remote regions, offers correct health guidance via symptoms
provided to the user.
This approach also reduces many of the limitations of the previous rule-based model since superior
algorithms combined with more complete data helped in flexible adaptation. It's based on data
privacy and fairness, thereby offering responsible AI-driven health support. Its future prospects, like
continuous learning with electronic health records, ensure an increased level of precision and
relevance for personalized medicine.
THANK
YOU