Spaces:
Sleeping
Sleeping
File size: 9,118 Bytes
742cd9a | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 | """
Pydantic schemas for request and response validation.
Every field that enters or leaves the API is defined here.
Pydantic validates types, ranges, and required fields automatically.
FastAPI returns a 422 error with clear messages if validation fails.
"""
from pydantic import BaseModel, Field, field_validator
from typing import Optional
from enum import Enum
class RiskCategory(str, Enum):
"""
Clinical risk categories with intervention thresholds.
Thresholds based on clinical literature and LACE index mapping.
"""
LOW = "LOW"
MEDIUM = "MEDIUM"
HIGH = "HIGH"
class PatientFeatures(BaseModel):
"""
Input schema — 31 engineered features for readmission prediction.
All features must be available at the point of patient discharge.
Field descriptions explain the clinical meaning of each input.
Min/max constraints catch impossible or clearly erroneous values.
"""
# --- Encounter features ---
los_days: float = Field(
...,
ge=1, le=365,
description="Length of stay in days (minimum 1 for inpatient)"
)
los_days_log: float = Field(
...,
ge=0,
description="Natural log of LOS (log1p transformed)"
)
admission_month: int = Field(
...,
ge=1, le=12,
description="Month of admission (1=January, 12=December)"
)
admission_dow: int = Field(
...,
ge=0, le=6,
description="Day of week (0=Monday, 6=Sunday)"
)
is_emergency: int = Field(
...,
ge=0, le=1,
description="1 if admitted via emergency department, 0 otherwise"
)
# --- Prior admission history ---
n_admissions_prior_6m: int = Field(
...,
ge=0, le=50,
description="Number of inpatient admissions in prior 6 months"
)
n_admissions_prior_12m: int = Field(
...,
ge=0, le=100,
description="Number of inpatient admissions in prior 12 months"
)
n_ed_visits_prior_6m: int = Field(
...,
ge=0, le=50,
description="Number of ED visits in prior 6 months (E in LACE)"
)
days_since_last_admission: float = Field(
...,
ge=0,
description="Days since previous admission. 999 if no prior admission."
)
has_prior_admission: int = Field(
...,
ge=0, le=1,
description="1 if patient has any prior inpatient admission"
)
# --- Comorbidity features ---
charlson_score: int = Field(
...,
ge=0, le=30,
description="Charlson Comorbidity Index score (C in LACE)"
)
n_active_conditions: int = Field(
...,
ge=0, le=200,
description="Total number of active clinical conditions"
)
has_heart_failure: int = Field(
..., ge=0, le=1,
description="1 if patient has heart failure diagnosis"
)
has_diabetes: int = Field(
..., ge=0, le=1,
description="1 if patient has diabetes mellitus diagnosis"
)
has_diabetes_complex: int = Field(
..., ge=0, le=1,
description="1 if patient has diabetes with complications"
)
has_copd: int = Field(
..., ge=0, le=1,
description="1 if patient has COPD diagnosis"
)
has_ckd: int = Field(
..., ge=0, le=1,
description="1 if patient has chronic kidney disease"
)
has_mi: int = Field(
..., ge=0, le=1,
description="1 if patient has history of myocardial infarction"
)
has_cancer: int = Field(
..., ge=0, le=1,
description="1 if patient has active cancer diagnosis"
)
has_dementia: int = Field(
..., ge=0, le=1,
description="1 if patient has dementia diagnosis"
)
has_cerebrovascular: int = Field(
..., ge=0, le=1,
description="1 if patient has cerebrovascular disease"
)
has_pvd: int = Field(
..., ge=0, le=1,
description="1 if patient has peripheral vascular disease"
)
# --- Medication features ---
n_medications_capped: float = Field(
...,
ge=0, le=520,
description="Active medication count (capped at 99th percentile)"
)
is_high_polypharmacy: int = Field(
..., ge=0, le=1,
description="1 if patient is on more than 10 medications"
)
has_insulin: int = Field(
..., ge=0, le=1,
description="1 if patient is prescribed insulin"
)
has_anticoagulant: int = Field(
..., ge=0, le=1,
description="1 if patient is prescribed anticoagulant therapy"
)
has_diuretic: int = Field(
..., ge=0, le=1,
description="1 if patient is prescribed diuretic therapy"
)
has_ace_inhibitor: int = Field(
..., ge=0, le=1,
description="1 if patient is prescribed ACE inhibitor"
)
# --- Demographic features ---
age_at_admission: float = Field(
...,
ge=18, le=120,
description="Patient age at time of admission (adults only)"
)
gender_male: int = Field(
..., ge=0, le=1,
description="1 if patient is male, 0 if female"
)
income: float = Field(
...,
ge=0,
description="Annual household income (socioeconomic proxy)"
)
@field_validator('los_days_log')
@classmethod
def validate_log_consistent(cls, v, info):
"""Verify log-transformed LOS is consistent with raw LOS."""
import numpy as np
if 'los_days' in info.data:
expected = np.log1p(info.data['los_days'])
if abs(v - expected) > 0.1:
raise ValueError(
f"los_days_log ({v:.3f}) inconsistent with "
f"los_days ({info.data['los_days']}) — "
f"expected ~{expected:.3f}"
)
return v
model_config = {
"json_schema_extra": {
"example": {
"los_days": 6.0,
"los_days_log": 1.946,
"admission_month": 2,
"admission_dow": 1,
"is_emergency": 0,
"n_admissions_prior_6m": 3,
"n_admissions_prior_12m": 3,
"n_ed_visits_prior_6m": 1,
"days_since_last_admission": 28.0,
"has_prior_admission": 1,
"charlson_score": 2,
"n_active_conditions": 29,
"has_heart_failure": 0,
"has_diabetes": 0,
"has_diabetes_complex": 0,
"has_copd": 0,
"has_ckd": 1,
"has_mi": 0,
"has_cancer": 1,
"has_dementia": 0,
"has_cerebrovascular": 0,
"has_pvd": 0,
"n_medications_capped": 47.0,
"is_high_polypharmacy": 1,
"has_insulin": 0,
"has_anticoagulant": 1,
"has_diuretic": 1,
"has_ace_inhibitor": 0,
"age_at_admission": 66.6,
"gender_male": 1,
"income": 125393.0
}
}
}
class RiskFactor(BaseModel):
"""A single SHAP-based risk factor contribution."""
feature: str = Field(..., description="Feature name")
value: float = Field(..., description="Actual feature value")
impact: float = Field(..., description="SHAP contribution to prediction")
direction: str = Field(..., description="'increases' or 'decreases' risk")
description: str = Field(..., description="Clinical plain-English explanation")
class PredictionResponse(BaseModel):
"""
Full prediction response returned to the clinician or EHR system.
Designed to be actionable — not just a number, but a clinical narrative
with the reasoning behind the score and a recommended action.
"""
readmission_risk: float = Field(..., description="Probability 0.0–1.0")
risk_percent: float = Field(..., description="Probability as percentage")
risk_category: RiskCategory = Field(..., description="LOW / MEDIUM / HIGH")
risk_score_display: str = Field(..., description="Human-readable risk display")
top_risk_factors: list[RiskFactor] = Field(..., description="Top SHAP contributors")
recommendation: str = Field(..., description="Clinical action recommendation")
model_version: str = Field(..., description="Model version identifier")
disclaimer: str = Field(
default="This prediction is a clinical decision support tool. "
"It does not replace clinical judgement.",
description="Mandatory clinical disclaimer"
)
class HealthResponse(BaseModel):
"""Health check response."""
status: str = Field(..., description="API status")
model_loaded: bool = Field(..., description="Whether model is loaded")
model_version: str = Field(..., description="Model version")
api_version: str = Field(..., description="API version") |