Phase 2: System Design

[FairyEvilMoon]

Design complex UML Diagrams:

• Use Case Diagrams for interactions between doctors, patients, and admins.
• Class Diagrams for system structure, focusing on entities like Patient, Doctor, Appointment,
• HealthMetrics, and Prescription.
• Sequence Diagrams for critical processes like real-time health monitoring and appointment
• scheduling.
• Component Diagrams to model the subsystems (patient portal, wearable integration,
• database, machine learning module).
• Design a Predictive Model for health risk analysis.

• Define the Database Schema for patient records, appointments, and health data.

  Deliverables:

- [x] UML Diagrams. - [x] Predictive Model Design. - [x] Database Schema

[FairyEvilMoon]

use diagram, class diagram, sequence diagram. if anyone is making a better one do post

PS. they are svg file, and because they are svg file that it displayed like trash so instead click on it or download it to have a better look at it

[FairyEvilMoon]

Component Diagram

[FairyEvilMoon]

Health Risk Predictive Model Design

1. Objective

Design a predictive model to analyze patient health data and predict potential health risks.

2. Input Features

• Age
• Gender
• Body Mass Index (BMI)
• Blood Pressure (Systolic and Diastolic)
• Heart Rate
• Cholesterol Levels (Total, HDL, LDL)
• Blood Glucose Level
• Smoking Status (Boolean)
• Alcohol Consumption (units per week)
• Physical Activity Level (hours per week)
• Family History of:
  • Heart Disease
  • Diabetes
  • Cancer

Previous Diagnoses

3. Output

The model will predict the probability of the patient developing the following conditions within the next 5 years:

• Cardiovascular Disease
• Type 2 Diabetes
• Hypertension

4. Model Architecture

We'll use an ensemble method combining multiple machine learning algorithms:

1. Random Forest Classifier

• Good for handling a mix of categorical and numerical data
• Can capture non-linear relationships
• Provides feature importance

2. Gradient Boosting Classifier (e.g., XGBoost)

• High performance and accuracy
• Can handle imbalanced datasets well

3. Logistic Regression

• Provides easily interpretable results
• Good for understanding the impact of individual features

5. Training Process

1. Data Preprocessing:

• Handle missing values
• Encode categorical variables
• Normalize numerical features

2. Feature Selection:

• Use correlation analysis and feature importance from Random Forest to select the most relevant features

3. Model Training:

• Split data into training (70%), validation (15%), and test (15%) sets
• Train each model (Random Forest, XGBoost, Logistic Regression) on the training data
• Use cross-validation to tune hyperparameters

4. Ensemble Method:

• Use soft voting to combine predictions from all three models
• Weight the predictions based on each model's performance on the validation set

5. Model Evaluation:

• Use metrics such as AUC-ROC, precision, recall, and F1-score
• Evaluate on the test set to assess generalization

6. Deployment and Monitoring

• Deploy the model as a microservice that can be called via API
• Implement logging to track model inputs and outputs
• Set up monitoring for model performance and data drift
• Plan for regular retraining (e.g., quarterly) to keep the model up-to-date

7. Interpretability

• Use SHAP (SHapley Additive exPlanations) values to explain individual predictions
• Provide feature importance plots to show which factors most influence the predictions

8. Ethical Considerations

• Ensure the training data is diverse and representative
• Regularly audit the model for bias, especially regarding protected characteristics like age, gender, and race
• Provide clear explanations of the model's limitations to healthcare providers

[FairyEvilMoon]

SHMS Database Schema

1. Users Table

sqlCopyCREATE TABLE Users (
    UserID INT PRIMARY KEY,
    Email VARCHAR(255) UNIQUE NOT NULL,
    PasswordHash VARCHAR(255) NOT NULL,
    FirstName VARCHAR(50) NOT NULL,
    LastName VARCHAR(50) NOT NULL,
    DateOfBirth DATE NOT NULL,
    Gender ENUM('Male', 'Female', 'Other') NOT NULL,
    PhoneNumber VARCHAR(20),
    UserType ENUM('Patient', 'Doctor', 'Admin') NOT NULL,
    CreatedAt TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    UpdatedAt TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP
);

2. Patients Table

sqlCopyCREATE TABLE Patients (
    PatientID INT PRIMARY KEY,
    UserID INT UNIQUE,
    EmergencyContact VARCHAR(255),
    InsuranceInfo TEXT,
    FOREIGN KEY (UserID) REFERENCES Users(UserID)
);

3. Doctors Table

sqlCopyCREATE TABLE Doctors (
    DoctorID INT PRIMARY KEY,
    UserID INT UNIQUE,
    Specialization VARCHAR(100),
    LicenseNumber VARCHAR(50) UNIQUE,
    FOREIGN KEY (UserID) REFERENCES Users(UserID)
);

4. Appointments Table

sqlCopyCREATE TABLE Appointments (
    AppointmentID INT PRIMARY KEY,
    PatientID INT,
    DoctorID INT,
    AppointmentDateTime DATETIME,
    Status ENUM('Scheduled', 'Completed', 'Cancelled') DEFAULT 'Scheduled',
    Notes TEXT,
    CreatedAt TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    UpdatedAt TIMESTAMP DEFAULT CURRENT_TIMESTAMP ON UPDATE CURRENT_TIMESTAMP,
    FOREIGN KEY (PatientID) REFERENCES Patients(PatientID),
    FOREIGN KEY (DoctorID) REFERENCES Doctors(DoctorID)
);

5. MedicalRecords Table

sqlCopyCREATE TABLE MedicalRecords (
    RecordID INT PRIMARY KEY,
    PatientID INT,
    DoctorID INT,
    Date DATE,
    Diagnosis TEXT,
    Treatment TEXT,
    Medications TEXT,
    Notes TEXT,
    FOREIGN KEY (PatientID) REFERENCES Patients(PatientID),
    FOREIGN KEY (DoctorID) REFERENCES Doctors(DoctorID)
);

6. HealthMetrics Table

sqlCopyCREATE TABLE HealthMetrics (
    MetricID INT PRIMARY KEY,
    PatientID INT,
    RecordedAt TIMESTAMP,
    HeartRate INT,
    BloodPressureSystolic INT,
    BloodPressureDiastolic INT,
    Temperature DECIMAL(3,1),
    OxygenSaturation INT,
    BloodGlucose INT,
    Weight DECIMAL(5,2),
    Steps INT,
    SleepHours DECIMAL(3,1),
    FOREIGN KEY (PatientID) REFERENCES Patients(PatientID)
);

7. Prescriptions Table

sqlCopyCREATE TABLE Prescriptions (
    PrescriptionID INT PRIMARY KEY,
    PatientID INT,
    DoctorID INT,
    MedicationName VARCHAR(100),
    Dosage VARCHAR(50),
    Frequency VARCHAR(50),
    StartDate DATE,
    EndDate DATE,
    Notes TEXT,
    FOREIGN KEY (PatientID) REFERENCES Patients(PatientID),
    FOREIGN KEY (DoctorID) REFERENCES Doctors(DoctorID)
);

8. HealthRiskPredictions Table

sqlCopyCREATE TABLE HealthRiskPredictions (
    PredictionID INT PRIMARY KEY,
    PatientID INT,
    PredictionDate DATE,
    CardiovascularDiseaseRisk DECIMAL(5,2),
    DiabetesRisk DECIMAL(5,2),
    HypertensionRisk DECIMAL(5,2),
    ModelVersion VARCHAR(20),
    FOREIGN KEY (PatientID) REFERENCES Patients(PatientID)
);

these data schemes might be modified in the future

[FairyEvilMoon]

- [ ] THE UML diagrams need to be looked at and reformated if needed

ok ignore everything at the top of this comment the work has to be restarted since the Dr. lowered the requirements therefore it should look much simpler then whatever gibberish was written before

[FairyEvilMoon]

• [ ] need to update and make new uml diagrams
• [ ] and new datascheme