Department of Computer Science And Enginerring

MINI PROJECT REVIEW

MEDICINE RECOMENDATION
SYSTEM:THE FUTURE HEALTH CARE

Team Members: Guided by,

GAUTHAMKUMAR J - 113121UG03027
ARUNRAJ K - 113121UG03013
Batch No: 14 Mrs.Hellan Sathya,
NAVEEN R - 113120UG03074 Date : 14.05.2024 Mtech Asst.Professor
Dept of CSE
 ABSTRACT
• The Drug Recommender System utilizes machine learning for sentiment analysis to
assist patients and medical professionals in making informed decisions about drug
prescriptions and treatments. This advanced strategy utilizes big data and sophisticated
analytics to analyze user sentiment and emotions in text data.

• By utilizing machine learning algorithms, the system extracts and categorizes

attitudes, providing valuable insights on medication effectiveness, adverse effects, and
patient satisfaction. Our approach combines sentiment analysis with drug
recommendation algorithms to optimize medicine selection, improve patient outcomes,
and reduce adverse reactions.

• Utilizing a large dataset of medication evaluations, we assess the efficacy of

the proposed method, demonstrating its ability to make accurate and
personalized recommendations. This approach has the potential to revolutionize
healthcare by providing patients and doctors with tools for informed decisions.
 OBJECTIVE
• To develop a software program capable of accurately detecting
Disease,Description,Precaution, Medications , Workouts, Diets for patient safety and
quality control.

• To develop an automated system for accurate and efficient identification of

Disease
using of Machine learining Techineques with Data sets.

• The system should also be user-friendly and easily accessible to both developers and
end-users.

• AI can serve as a tool to support healthcare professionals in their decision – making

procress. By providing quick access to comprehensive medication information. AI can
assist in selecting appropriate medications, identifying potential interactions or
contraindications, and ensuring proper dosing and administration.
 INTRODUCTION
1. Automated Disease Diagnosis and Precaution Recommender System:
1. This system leverages supervised machine learning to assist in disease diagnosis and recommend
necessary precautions.
2. It aims to substitute an initial stage of diagnosis by providing accurate predictions based on symptoms.
3. The approach involves two modules:
1. Module 1: Trains machine learning models using a dataset of disease symptoms, corresponding symptoms,
and precautions.
1. Algorithms applied include support vector machine, random forest, extra trees classifier,
logistic regression, multinomial naive Bayes, and decision tree.
2. Module 2: Interacts with the user (patient) via voice input. The patient describes illness symptoms using
a microphone, and the system predicts the disease and recommends precautions.
4. Real-time evaluation achieves an impressive accuracy of 99.9%.
2. Ensemble Learning for Disease Prediction:
1. Ensemble learning combines multiple classifiers to enhance performance.
2. It improves prediction accuracy compared to using a single classifier.
3. Early Prediction of Lifestyle Diseases using Machine Learning:
1. In this project, a machine learning application predicts the disease.
2. The goal is to raise individual health awareness and prompt timely care, potentially saving lives.
 EXISTING SYSTEM
•Food Recommendation: Providing dietary advice based on individual health
conditions.

•Drug Recommendation: Suggesting appropriate medications based on symptoms,

allergies, and patient history.

•Health Status Prediction: Predicting health outcomes based on historical data.

• Healthcare Service Recommendation: Guiding patients toward relevant

healthcare services.

• Healthcare Professional Recommendation: Matching patients with suitable

doctors or specialists.
 EXISTING BLOCK DIAGRAM

Data Cleaning Train test

Data set and Vectroization split
preprocressing

MEDICINE RECOMMENDATION Model

Buliding
SYSTEM PIPELINE

Prediction of Model Model

Medicine Selection Validation
 DISADVANTAGES
• Information Overload: Medical professionals face an increasing volume of drugs, tests,
and treatment recommendations daily. Deciding the best treatment based on
symptoms,test results, and medical history becomes challenging.

• Time-Consuming: Manual identification methods are time-consuming, especially when

dealing with a large number of pills.

• Subjectivity: Human identification can be subjective, as different individuals may

interpret pill characteristics differently.

• Human Error: Human-based identification methods are susceptible to errors, such as

misinterpretation of pill characteristics, fatigue, and oversight.
 PROPOSED SYSTEM
• Dataset Creation and Augmentation: Collect a diverse dataset of
Disease,Description,Precaution, Medications , Workouts, Diets including
different disease and its measures.

• Machine Learning Model: Implement a Convolutional Neural Network (CNN)

architecture to train and identification of Disease and its medicine.

• User-Friendly Interface: Develop a user-friendly web application that allows

users to enter the Disease of pills , diets ,precations for identification.

• Real-Time Identification: Enable real-time pill identification, where users can use to
enter the Disease for immediate identification.
 PROPOSED BLOCK DIAGRAM

Data Cleaning Train test Model

Data set and Vectroization
split Buliding
preprocressing

Prediction of
Diets Model
MEDICINE RECOMMENDATION SYSTEM PIPELINE
Validation

Prediction of
Workouts

Prediction of Prediction of Prediction of Prediction of Model

Medications Precaution Description Disease Selection
 ADVANTAGES
•Accuracy: The automated system will provide accurate prediction of Medicine
identification, reducing the risk of medication errors and enhancing patient safety.

•Efficiency: Users can quickly identify Medicine and its precautions without the need
for manual, time-consuming processes, making it suitable for busy healthcare settings.

•Consistency: The system will offer consistent results, eliminating the subjectivity
associated with human-based identification methods.

• Patient Empowerment: Patients become active participants in their healthcare journey.

Access to personalized recommendations empowers them to make informed decisions.

• Efficient Decision-Making: In a vast sea of medical information, these systems filter

relevant details.Physicians can quickly access evidence-based recommendations, saving
time and improving patient care.
 COMPARISON
Aspect Existing System Proposed System
High accuracy through deep
Accuracy Reliant on expertise
learning and computer vision

Limited scalability and time- Highly efficient, capable of real-

Efficiency
consuming time identification

Consistency Subjective, varying results Consistent and objective results

Enhanced ability to detect accurate

Decision Making Limited effectiveness
medications

11
 LIST OF MODULES
 Module 1: Trains machine learning models using a dataset of disease symptoms,
corresponding symptoms, and precautions.
 Algorithms applied include support vector machine, random forest, extra trees classifier,
logistic regression, multinomial naive Bayes, and decision tree.

 Module 2: Interacts with the user (patient) via voice input. The patient describes illness
symptoms using a microphone, and the system predicts the disease and recommends precautions.

 Module 3: Scikit-learn is built on top of other popular Python libraries like NumPy, pandas, and
Matplotlib.

 Module 4: supervised learning techniques, such as linear regression, decision trees, support vector
machines (SVM), and k-nearest neighbors (k-
NN).
 Module 5: Pickling is the process of converting a Python object hierarchy into a byte stream.When
you pickle an object, it gets transformed into a format that can be saved to disk or transmitted over a
network.
 LITERATURE SURVEY

A Survey on Medicine Identification Techniques Using Machine Learning:

MERIT: Provides an overview of various Machine Learning techniques used in pill
identification. The survey highlights the importance of understanding the basics before
implementing advanced techniques.
DEMERIT: May lack specific details on the latest technologies due to the broad scope.
Deep Learning Approaches for Symptoms Recognition: A Comprehensive Review:
MERIT: Focuses on deep learning techniques for Symptoms identification. Discusses the
advantages of using convolutional neural networks (CNNs) in this context.
DEMERIT: Might not cover traditional Machine Learning methods, potentially
neglecting their merits.

13
 LITERATURE SURVEY Contd.
A Fair and Safe Usage Drug Recommendation System in Medical Emergencies:
MERITS:Researchers proposed a system using a stacked artificial neural network (ANN)
model.
•The goal was to improve the fairness and safety of treatment for infectious diseases.
•The system recommends drugs based on a patient’s previous health profile, lifestyle, and habits.
•Achieved an impressive 97.5% accuracy in drug recommendations.
•Such a system could be especially useful during health emergencies.
•DEMRITS:May not delve into the scalability of the proposed system for a large variety of pills.
A Comparative Study on Pill Recognition: Traditional Methods versus Machine
Learning Approaches:
MERIT: Provides a comprehensive comparison between traditional computer vision methods
and machine learning approaches for pill identification.
DEMERIT: The study may not focus extensively on the latest advancements in either
traditional or machine learning methods, potentially missing out on cutting-edge
technologies.
 FLOW CHART
Organize
images into
Load
folders
Dataset according to
their labels
Compile
Model
Disease Data Adam
Data Pre optimizer Pre
Generator Save Load
Categorical Train process Make Display
processing based on Trained Trained Input
cross Model prediction Results
Symptoms Model Model
entropy image
loss
accuracy
Transfer metric
Define Learning Load
Model pre-trained
Architecture model without
top layers

15
 ALGORITHM
USAGE
Data Preprocessing - Performs various data augmentation techniques, such as rescaling,
shearing, zooming, and horizontal flipping, to artificially increase the diversity of the
training set and improve the model's ability to generalize to new data.
Deep Learning - Researchers have applied various deep learning (DL) models to patients’ EHRs
recommend prescriptions.DL-based
to medication recommendation (MR) models analyze patient data and
suggest appropriate medications.
Random Forest -Is an ensemble learning method that combines multiple decision trees.It
accuracy by reducing overfitting and handling noisy data .In medicine recommendation,
improves
Random
Forest considers various patient attributes to suggest appropriate
medication
Personalized Medicine Recommendations- The goal of personalized medicine recommendation
systems is to tailor drug suggestions to individual patients.These systems consider medical
conditions,
symptoms, allergies, and other relevant factors.Machine learning models, such as neural
collaborative filtering, and hybrid approaches, contribute to personalized
networks,
recommendations

16
 LIBRARIES
USED
Tensorflow – Machine learning library for building and training neural network models.

Flask – Is a API of python that allows us to create a web application

Numpy – NumPy can be used to perform a wide variety of mathematical operations on arrays.

Pandas - Pandas is a Python library used for working with data sets. It has functions for
analyzing, cleaning, exploring, and manipulating data.

Pickle– Python's Pickle module is a popular format used to serialize and deserialize data types.

17
 HARDWARE
REQUIREMENTS
Operating System : Windows 11
Processor : Intel Core i5 11th Gen
Video card : NVIDIA GTX GEFORCE 1050TI
Memory : 8GB RAM
Resolution : 1920X1080
Storage (Hard Drives) : SSDs (Solid State Drives) are faster than traditional
HDDs (Hard Disk Drives) and are recommended for better
performance.

18
 SOFTWARE REQUIREMENTS

Software tool : PyCharm / Visual Studio

Modules : TensorFlow, NumPy, Pandas, Flask, Pickle,

Scikit-learn.

Development Environment : Use the integrated Environment IDE

Jupiter Notebook and

Frameworks : PyTorch: Popular deep learning frameworks.

Keras: A high-level API for building neural

networks.
 IMPLEMENTATION
There are programs in our project:

Main.py – Program to build the model using svc/pkl .

Index.html – To create a index webpage .

Developer.html – Its shows the Detail about the Developer Info.

Blog.html – Its show about the blog detail
Contact.html – Its show the Conctact deatails about the Website and Developer
About.html – Its Shows about the webpage.
MAIN.PY
 EVALUATION

ACCURACY AND LOSS

CONFUSION MATRIX

The trained model has an accuracy of 93%.

The overall accuracy is 25%.

This can be increased by:

• Collecting more data
• Using advanced architectures

29
OUTPUT
 FUTURE ENHANCEMENTS

Data Enrichment - Collect and incorporate a more extensive and diverse dataset of
Disease to enhance the model's recognition capabilities.

Web Application Interaction - Develop a user-friendly web application to make the

Medicine identification system accessible to a broader audience.

Database Optimization - Explore ways to optimize database for efficient storage and
retrieval of pill-related information.

33
 FUTURE ENHANCEMENTS
Contd..

Model Architecture Upgrade - Replace the Machine learning architecture with a more
sophisticated CNN model, considering architectures like ResNet or EfficientNet for
improved accuracy.

34
 CONCLUSION
The proposed automated Medicine identification system, powered by Machine Learning
and Deep Learning, offers a highly accurate and efficient solution for identifying Indian
Medicine based on visual characteristics.

This system addresses the limitations of existing manual methods, providing consistent and
objective results, improved scalability, and enhanced detection capabilities.

By revolutionizing Medicine identification, this technology contributes to patient safety,

pharmaceutical quality control, and accessibility, offering a valuable tool for both
healthcare professionals and individuals.

35
 REFERENCES
•P. Chotivatunyu and N. Hnoohom, "Medicine Identification System on Mobile Devices for the
elderly," 2020 15th International Joint Symposium on Artificial Intelligence and Natural
Language Processing (iSAI-NLP), 2020, pp. 1-6, doi: 10.1109/iSAI-
NLP51646.2020.9376837.

•Sangamithra, B. "Personalized Ranking Mechanism Using Yandex Dataset on Machine Learning

Approaches." Proceedings of the International Conference on Cognitive and Intelligent
Computing. Springer, Singapore, 2022.

•Ting, HW., Chung, SL., Chen, CF. et al. A drug identification model developed using deep
learning technologies: experience of a medical center in Taiwan. BMC Health Serv Res 20, 312
(2020).

•Rizwan Patan, Suresh Kallam, and Mohamed Yasin Noor Mohamed. "Segmentation of nuclei
in histopathology images using fully convolutional deep neural architecture." In 2020
International Conference on computing and information technology (ICCIT-1441), pp. 1-7.
IEEE, 2020. 36
THANK YOU