Patient Features - Diabetes Prediction ML

Overview

The diabetes prediction model analyzes 8 patient characteristics to assess diabetes risk. This page provides detailed medical context for each feature, helping healthcare professionals understand what the model considers when making predictions.

Total Features: 8 (2 categorical, 6 numeric)Clinical Basis: All features are established diabetes risk factors in medical literature

Feature Summary Table

Feature	Type	Clinical Significance	Normal Range	High Risk
Gender	Categorical	Metabolic differences	N/A	Male (slightly)
Age	Numeric	Risk increases with age	N/A	>45 years
Hypertension	Binary	Insulin resistance marker	0 (no)	1 (yes)
Heart Disease	Binary	Metabolic syndrome indicator	0 (no)	1 (yes)
Smoking History	Categorical	Inflammatory effects	never	current
BMI	Numeric	Obesity indicator	18.5-24.9	≥30
HbA1c Level	Numeric	Long-term glucose control	<5.7%	≥6.5%
Blood Glucose	Numeric	Current glucose level	<100 mg/dL	≥126 mg/dL

1. Gender

Description

Biological sex of the patient.

Medical Significance
Data Format
Example Cases

Why Gender Matters for Diabetes:

Hormonal Differences: Estrogen and testosterone affect insulin sensitivity
Body Composition: Men tend to accumulate visceral fat (higher diabetes risk)
Prevalence: Slightly higher diabetes rates in men in some populations
Complications: Women with diabetes have higher cardiovascular mortality

Research Notes:

Type 2 diabetes is ~1.5x more common in men
Women with PCOS (polycystic ovary syndrome) have elevated risk
Gestational diabetes increases future type 2 risk in women

Categorical Values:

"Female" | "Male" | "Other"

Encoded Values:

{
  "Female": 0,
  "Male": 1,
  "Other": 2
}

Distribution (typical):

Female: ~50%
Male: ~50%
Other: <1%

{
  "gender": "Female",
  "age": 45,
  // More features...
}

{
  "gender": "Male",
  "age": 55,
  // More features...
}

2. Age

Description

Patient’s age in years.

Clinical Significance

Risk Progression

Diabetes risk increases dramatically after age 45, with peak incidence in 65-79 age group

Physiological Changes

Aging causes decreased insulin secretion, increased insulin resistance, and reduced beta-cell function

Age Range    | Diabetes Risk | Contributing Factors
-------------|---------------|---------------------
< 30 years   | Low           | Peak metabolic function
30-44 years  | Moderate      | Lifestyle factors emerge
45-64 years  | High          | Insulin resistance increases
65+ years    | Very High     | Multiple comorbidities

Medical Context

Why Age Increases Diabetes Risk

Beta-Cell Decline: Pancreatic beta cells produce less insulin with age
Increased Visceral Fat: Abdominal fat accumulation worsens with age
Mitochondrial Dysfunction: Cellular energy metabolism becomes less efficient
Chronic Inflammation: Age-related inflammation affects insulin signaling
Sarcopenia: Muscle loss reduces glucose disposal capacity

Data Format

{
  "age": 36,
  // Integer, typically 0-120
}

Example Risk Profiles

Low Risk (Young)
Moderate Risk (Middle Age)
High Risk (Elderly)

{
  "age": 28,
  "bmi": 22.0,
  "HbA1c_level": 4.9
}

Assessment: Young age is protective even with other risk factors present

{
  "age": 45,
  "bmi": 28.5,
  "HbA1c_level": 5.8
}

Assessment: Age 45+ is a screening threshold in clinical guidelines

{
  "age": 70,
  "bmi": 30.0,
  "HbA1c_level": 6.0
}

Assessment: Elderly patients need careful monitoring even with borderline values

3. Hypertension

Description

Indicates whether the patient has been diagnosed with high blood pressure.

Clinical Significance

Hypertension and diabetes frequently co-occur in metabolic syndrome:

Shared Pathophysiology

Both conditions involve:

Insulin resistance
Endothelial dysfunction
Chronic inflammation
Sympathetic nervous system overactivity

Bidirectional Risk

Hypertension increases diabetes risk by 2-3x
Diabetes increases hypertension risk
Together, they amplify cardiovascular complications

Clinical Impact

Patients with both conditions face:

3-4x higher stroke risk
2-3x higher heart attack risk
Accelerated kidney damage
Increased retinopathy (eye damage)

Medical Definition

Hypertension Diagnosis:

Systolic BP ≥ 140 mmHg, OR
Diastolic BP ≥ 90 mmHg
Measured on multiple occasions

Data Format

{
  "hypertension": 0  // 0 = No, 1 = Yes
}

Binary Values:

0: No hypertension diagnosis
1: Has hypertension

Prevalence

Co-occurrence: ~75% of people with diabetes also have hypertensionIn Dataset: Approximately 7-10% of patients have hypertension=1

4. Heart Disease

Description

Indicates whether the patient has been diagnosed with cardiovascular disease.

Clinical Significance

Heart disease and diabetes share common risk factors and mechanisms:

Atherosclerosis

Both conditions accelerate plaque buildup in arteries through oxidative stress and inflammation

Metabolic Syndrome

Heart disease, diabetes, hypertension, and obesity cluster together in metabolic syndrome

Relationship with Diabetes

Why Heart Disease Predicts Diabetes

Shared Etiology: Both stem from insulin resistance and metabolic dysfunction
Inflammatory State: Cardiovascular inflammation impairs insulin signaling
Endothelial Dysfunction: Damaged blood vessel lining affects glucose metabolism
Medication Effects: Some cardiac medications (beta-blockers, thiazides) can worsen glucose control

Data Format

{
  "heart_disease": 0  // 0 = No, 1 = Yes
}

Binary Values:

0: No heart disease diagnosis
1: Has heart disease (CAD, MI, CHF, etc.)

Clinical Implications

High-Risk Combination: Patients with both diabetes and heart disease require aggressive management:

Tight blood sugar control (HbA1c < 7%)
Blood pressure < 130/80 mmHg
LDL cholesterol < 70 mg/dL
Antiplatelet therapy
ACE inhibitors or ARBs

5. Smoking History

Description

Patient’s smoking status and history.

Clinical Significance

Smoking affects diabetes risk through multiple mechanisms:

Insulin Resistance

Nicotine and other toxins impair insulin signaling, increasing blood glucose

Inflammation

Smoking triggers chronic systemic inflammation, a key driver of metabolic disease

Abdominal Fat

Smokers tend to accumulate visceral fat, the most metabolically harmful type

Beta-Cell Toxicity

Smoking may directly damage insulin-producing pancreatic cells

Data Format

{
  "smoking_history": "current"
  // Options: "No Info", "current", "ever", "former", "never", "not current"
}

Risk Quantification

Relative Diabetes Risk by Smoking Status:

Smoking Status    | Relative Risk | Compared to Never Smokers
------------------|---------------|---------------------------
Never             | 1.0x          | Baseline
Former (quit >10y)| 1.1x          | Nearly baseline
Former (quit <10y)| 1.2x          | Slightly elevated
Current (light)   | 1.3x          | Moderate increase
Current (heavy)   | 1.5x          | Significant increase

6. BMI (Body Mass Index)

Description

A measure of body fat based on height and weight: BMI = weight (kg) / height² (m²)

Clinical Significance

Most Important Modifiable Risk Factor: Obesity is the single strongest predictor of type 2 diabetes, with BMI ≥30 increasing risk by 7-10x.

BMI Categories

Classification
Pathophysiology
Clinical Thresholds

BMI Range	Category	Diabetes Risk
< 18.5	Underweight	Low
18.5 - 24.9	Normal Weight	Baseline (1x)
25.0 - 29.9	Overweight	2-3x
30.0 - 34.9	Obese (Class I)	5-7x
35.0 - 39.9	Obese (Class II)	7-10x
≥ 40.0	Obese (Class III)	10-12x

Data Format

{
  "bmi": 32.27
  // Float, typical range 10-95
}

Example Calculations:

# BMI = weight (kg) / height² (m²)

# Example 1: Normal weight
weight = 70  # kg
height = 1.75  # meters
bmi = 70 / (1.75 ** 2) = 22.86
# Classification: Normal

# Example 2: Obese
weight = 95  # kg
height = 1.70  # meters
bmi = 95 / (1.70 ** 2) = 32.87
# Classification: Obese Class I

Example Profiles

{
  "bmi": 22.5,
  "age": 30,
  "HbA1c_level": 5.0
}
// Prediction: Likely no diabetes

7. HbA1c Level (Hemoglobin A1c)

Description

Average blood glucose level over the past 2-3 months, expressed as a percentage.

Clinical Significance

Gold Standard: HbA1c is the primary diagnostic test for diabetes and the best indicator of long-term glucose control.

How HbA1c Works

Biochemistry

Glycation Process:

Glucose in blood binds to hemoglobin in red blood cells
Higher blood glucose → more glucose-hemoglobin binding (glycation)
Red blood cells live ~120 days, so HbA1c reflects 2-3 month average
Measured as percentage of hemoglobin that’s glycated

Advantages:

Not affected by recent meals (fasting not required)
Stable day-to-day
Reflects long-term control
Strong predictor of complications

Diagnostic Thresholds

Classification
Risk Stratification
Treatment Targets

HbA1c Level	Category	Interpretation
< 5.7%	Normal	Healthy glucose metabolism
5.7% - 6.4%	Prediabetes	Impaired glucose tolerance; reversible
≥ 6.5%	Diabetes	Diagnostic for diabetes (if confirmed)
≥ 8.0%	Poorly Controlled	Urgent intervention needed
≥ 9.0%	Very Poor Control	High complication risk

HbA1c and Diabetes Risk:

HbA1c    | 5-Year Diabetes Risk
---------|----------------------
< 5.0%   | < 1%
5.0-5.4% | 5%
5.5-5.9% | 10%
6.0-6.4% | 25-50%
≥ 6.5%   | Diabetes (diagnosed)

Data Format

{
  "HbA1c_level": 6.2
  // Float, typically 3.5-15.0
  // Expressed as percentage
}

Conversion to Average Glucose

Estimated Average Glucose (eAG):

eAG (mg/dL) = 28.7 × HbA1c - 46.7

Examples:
HbA1c 5.0% → eAG = 97 mg/dL (normal)
HbA1c 6.0% → eAG = 126 mg/dL (prediabetic)
HbA1c 7.0% → eAG = 154 mg/dL (diabetic)
HbA1c 8.0% → eAG = 183 mg/dL (poorly controlled)

Example Risk Profiles

{
  "HbA1c_level": 5.0,
  "blood_glucose_level": 95
}
// Assessment: Healthy glucose metabolism

Critical Note: HbA1c ≥ 6.5% is diagnostic for diabetes when confirmed with repeat testing. Patients with values in this range should be immediately flagged for clinical evaluation.

8. Blood Glucose Level

Description

Current blood glucose (sugar) concentration, measured in mg/dL (milligrams per deciliter).

Clinical Significance

Snapshot Measurement

Unlike HbA1c (2-3 month average), blood glucose shows current glucose level at time of measurement

State-Dependent

Interpretation depends on when measured: fasting, post-meal, random, etc.

Diagnostic Thresholds

Fasting Glucose
Random Glucose
Oral Glucose Tolerance Test

Measured after 8+ hours without food:

Glucose Level	Category	Interpretation
< 100 mg/dL	Normal	Healthy glucose regulation
100-125 mg/dL	Prediabetes	Impaired fasting glucose
≥ 126 mg/dL	Diabetes	Diagnostic if confirmed

Dataset Context: The dataset doesn’t specify fasting status, so interpretation requires caution.

Measured at any time, regardless of meals:

Glucose Level	Category	Interpretation
< 140 mg/dL	Usually Normal	But depends on meal timing
140-199 mg/dL	Elevated	Warrants further testing
≥ 200 mg/dL	Diabetes	Diagnostic if symptomatic

Note: Random glucose ≥ 200 mg/dL with symptoms (thirst, urination, weight loss) is diagnostic for diabetes.

Glucose Level	Category	Interpretation
< 140 mg/dL	Normal	Normal glucose tolerance
140-199 mg/dL	Prediabetes	Impaired glucose tolerance
≥ 200 mg/dL	Diabetes	Diagnostic

Data Format

{
  "blood_glucose_level": 220
  // Integer, typically 60-400 mg/dL
}

Clinical Ranges

Glucose Range     | Clinical Meaning           | Action
------------------|----------------------------|------------------
< 70 mg/dL        | Hypoglycemia               | Immediate treatment
70-99 mg/dL       | Normal (fasting)           | No action
100-125 mg/dL     | Prediabetes (fasting)      | Lifestyle counseling
126-139 mg/dL     | Diabetes (fasting)         | Confirm diagnosis
140-199 mg/dL     | Elevated (random)          | Further testing
200-250 mg/dL     | Hyperglycemia              | Medical evaluation
> 250 mg/dL       | Severe hyperglycemia       | Urgent care
> 600 mg/dL       | Hyperosmolar state         | Emergency

Relationship with HbA1c

Blood glucose and HbA1c should correlate:

{
  "HbA1c_level": 5.0,
  "blood_glucose_level": 90
}
// Both normal - consistent story

Example Patient Cases

Case 1: Normal
Case 2: Prediabetic
Case 3: Diabetic

{
  "blood_glucose_level": 85,
  "HbA1c_level": 4.9,
  "bmi": 22.0,
  "age": 30
}

Assessment: Healthy glucose levels, low diabetes risk

{
  "blood_glucose_level": 115,
  "HbA1c_level": 5.9,
  "bmi": 28.5,
  "age": 48
}

Assessment: Prediabetic range; intensive lifestyle intervention recommended

{
  "blood_glucose_level": 220,
  "HbA1c_level": 8.2,
  "bmi": 34.0,
  "age": 55
}

Assessment: Poorly controlled diabetes; medication adjustment needed

Feature Interactions

The model considers combinations of features:

High-Risk Combinations

Metabolic Syndrome Profile:

{
  "bmi": 32,              // Obese
  "hypertension": 1,       // High BP
  "blood_glucose_level": 120,  // Elevated
  "age": 55               // Middle age
}

All four components of metabolic syndrome present → Very high diabetes risk

Protective Combinations

Low-Risk Profile:

{
  "age": 28,
  "bmi": 21.5,
  "smoking_history": "never",
  "hypertension": 0,
  "HbA1c_level": 4.8
}

Young, healthy weight, non-smoker with excellent glucose control

Clinical Applications

Screening Decisions

Use feature values to determine screening intervals:

if age >= 45 or (age >= 35 and bmi >= 25):
    screen_now = True
    interval = "Every 3 years if normal"
else:
    screen_now = False
    interval = "Wait until age 45"

Intervention Priorities

Focus interventions based on modifiable features:

BMI: Weight loss is first-line intervention
Smoking: Cessation reduces risk and complications
Blood Pressure: Control hypertension if present
Glucose: Lifestyle changes to lower HbA1c

Non-modifiable (focus on monitoring):

Age
Gender
Heart disease history

Summary: Feature Importance

Based on medical literature and model feature importance:

Rank | Feature            | Modifiable | Impact
-----|--------------------|-----------|---------
  | HbA1c_level        | Yes       | Very High
  | blood_glucose      | Yes       | Very High
  | bmi                | Yes       | High
  | age                | No        | High
  | hypertension       | Partial   | Moderate
  | smoking_history    | Yes       | Moderate
  | heart_disease      | Partial   | Moderate
  | gender             | No        | Low

Next Steps

Model Architecture

Learn how these features are used in the RandomForest model

Data Preprocessing

Understand how features are encoded and scaled

Dataset Overview

See the full dataset structure and distribution

Quick Start

Start making predictions with these features

Overview

Getting Started

Core Concepts

Deployment

Documentation Index

​Overview

​Feature Summary Table

​1. Gender

​Description

​2. Age

​Description

​Clinical Significance

Risk Progression

Physiological Changes

​Age-Related Risk Factors

​Medical Context

​Data Format

​Example Risk Profiles

​3. Hypertension

​Description

​Clinical Significance

​Medical Definition

​Data Format

​Prevalence

​4. Heart Disease

​Description

​Clinical Significance

Atherosclerosis

Metabolic Syndrome

​Relationship with Diabetes

​Data Format

​Clinical Implications

​5. Smoking History

​Description

​Clinical Significance

​Categories

​Data Format

​Risk Quantification

​6. BMI (Body Mass Index)

​Description

​Clinical Significance

​BMI Categories

​Data Format

​Example Profiles

​7. HbA1c Level (Hemoglobin A1c)

​Description

​Clinical Significance

​How HbA1c Works

​Diagnostic Thresholds

​Data Format

​Conversion to Average Glucose

​Example Risk Profiles

​8. Blood Glucose Level

​Description

​Clinical Significance

Snapshot Measurement

State-Dependent

​Diagnostic Thresholds

​Data Format

​Clinical Ranges

​Relationship with HbA1c

​Example Patient Cases

​Feature Interactions

​High-Risk Combinations

​Protective Combinations

​Clinical Applications

​Screening Decisions

​Intervention Priorities

​Summary: Feature Importance

​Next Steps

Model Architecture

Data Preprocessing

Dataset Overview

Quick Start

Build docs developers (and LLMs) love

Overview

Feature Summary Table

1. Gender

Description

2. Age

Description

Clinical Significance

Age-Related Risk Factors

Medical Context

Data Format

Example Risk Profiles

3. Hypertension

Description

Clinical Significance

Medical Definition

Data Format

Prevalence

4. Heart Disease

Description

Clinical Significance

Relationship with Diabetes

Data Format

Clinical Implications

5. Smoking History

Description

Clinical Significance

Categories

Data Format

Risk Quantification

6. BMI (Body Mass Index)

Description

Clinical Significance

BMI Categories

Data Format

Example Profiles

7. HbA1c Level (Hemoglobin A1c)

Description

Clinical Significance

How HbA1c Works

Diagnostic Thresholds

Data Format

Conversion to Average Glucose

Example Risk Profiles

8. Blood Glucose Level

Description

Clinical Significance

Diagnostic Thresholds

Data Format

Clinical Ranges

Relationship with HbA1c

Example Patient Cases

Feature Interactions

High-Risk Combinations

Protective Combinations

Clinical Applications

Screening Decisions

Intervention Priorities

Summary: Feature Importance

Next Steps