iints-sdk-python35 1.3.1__py3-none-any.whl → 1.4.0__py3-none-any.whl

iints/__init__.py

@@ -11,7 +11,7 @@ except ImportError:  # pragma: no cover - Python < 3.8 fallback
 try:
     __version__ = version("iints-sdk-python35")
 except PackageNotFoundError:  # pragma: no cover - source tree fallback
-    __version__ = "1.3.1"
+    __version__ = "1.4.0"
 
 # Note to developers: this SDK is currently maintained by a single author.
 # Please report bugs via GitHub issues and feel free to contribute fixes via PRs.

iints/ai/backends/ollama.py

@@ -2,9 +2,11 @@ from __future__ import annotations
 
 import json
 import os
+from ipaddress import ip_address
 from http.client import IncompleteRead, RemoteDisconnected
 from time import sleep
 from urllib import error, request
+from urllib.parse import urlparse
 
 
 DEFAULT_OLLAMA_HOST = "http://127.0.0.1:11434"
@@ -35,10 +37,38 @@ class OllamaBackend:
         timeout_seconds: float = 120.0,
     ) -> None:
         self.model_name = model_name
-        self.base_url = (base_url or os.getenv("OLLAMA_HOST") or DEFAULT_OLLAMA_HOST).rstrip("/")
+        raw_base_url = base_url or os.getenv("OLLAMA_HOST") or DEFAULT_OLLAMA_HOST
+        self.base_url = self._validate_base_url(raw_base_url)
         self.timeout_seconds = timeout_seconds
         self.resolved_model_name: str | None = None
 
+    @staticmethod
+    def _is_loopback_host(hostname: str) -> bool:
+        if hostname == "localhost":
+            return True
+        try:
+            return ip_address(hostname).is_loopback
+        except ValueError:
+            return False
+
+    @classmethod
+    def _validate_base_url(cls, raw_base_url: str) -> str:
+        parsed = urlparse(raw_base_url)
+        if parsed.scheme not in {"http", "https"}:
+            raise ValueError(
+                "OLLAMA_HOST/base_url must use http or https. Other URL schemes are blocked."
+            )
+        if not parsed.hostname:
+            raise ValueError("OLLAMA_HOST/base_url must include a hostname.")
+        if parsed.path not in {"", "/"}:
+            raise ValueError("OLLAMA_HOST/base_url must not include a path component.")
+        if not cls._is_loopback_host(parsed.hostname) and os.getenv("IINTS_ALLOW_REMOTE_OLLAMA") != "1":
+            raise ValueError(
+                "Remote Ollama endpoints are disabled by default. "
+                "Use localhost/127.0.0.1 or set IINTS_ALLOW_REMOTE_OLLAMA=1 explicitly."
+            )
+        return raw_base_url.rstrip("/")
+
     def _pull_hint(self) -> str:
         return f"ollama pull {self.model_name}"
 
@@ -102,7 +132,7 @@ class OllamaBackend:
             headers["Content-Type"] = "application/json"
         req = request.Request(url, data=body, headers=headers, method=method)
         try:
-            with request.urlopen(req, timeout=self.timeout_seconds) as response:
+            with request.urlopen(req, timeout=self.timeout_seconds) as response:  # nosec B310 - base_url is scheme/host validated
                 text = response.read().decode("utf-8")
         except error.HTTPError as exc:
             detail = exc.read().decode("utf-8", errors="replace").strip()

iints/ai/mdmp_guard.py

@@ -14,9 +14,8 @@ def _load_mdmp_verifier() -> type[Any]:
         module = importlib.import_module("mdmp_core")
     except Exception as exc:
         raise ImportError(
-            "MDMP verification requires the optional standalone package.\n"
-            "Install with: pip install iints-sdk-python35[mdmp]\n"
-            "or: pip install 'mdmp-protocol>=0.3.0'"
+            "MDMP verification requires the bundled MDMP crypto support.\n"
+            "Install with: pip install iints-sdk-python35[mdmp]"
         ) from exc
     verifier_cls = getattr(module, "MDMPVerifier", None)
     if verifier_cls is None:

iints/ai/prepare.py

@@ -245,7 +245,7 @@ def _load_mdmp_signer_tools() -> tuple[type[Any], Any]:
         module = importlib.import_module("mdmp_core")
     except Exception as exc:
         raise ImportError(
-            "Local AI certification requires the optional standalone MDMP package.\n"
+            "Local AI certification requires the bundled MDMP crypto support.\n"
             "Install with: pip install 'iints-sdk-python35[mdmp]'"
         ) from exc
 

iints/analysis/booth_demo.py

@@ -233,7 +233,7 @@ def _build_commands_markdown(
         "# Booth Demo Commands\n\n"
         "## Showable live demo script\n\n"
         "```bash\n"
-        f"PYTHONPATH=src python3 {example_script} --output-dir {output_dir}\n"
+        f"python3 {example_script} --output-dir {output_dir}\n"
         "```\n\n"
         "## Run from source tree\n\n"
         "```bash\n"
@@ -264,6 +264,8 @@ def _build_live_demo_script_text(
         "1. WHAT CODE TO SHOW FIRST\n"
         "- Show examples/demos/07_live_stage_demo.py first.\n"
         "  Reason: the top of that file exposes the patient profile, output folder, duration, and seed on one screen.\n"
+        "- Point out the visible SDK feature calls in that script:\n"
+        "  run_full(...), generate_results_poster(...), and prepare_ai_ready_artifacts(...).\n"
         "- Point out that you can swap PATIENT_CONFIG to another packaged profile such as patient_559_config or clinic_safe_hypo_prone.\n"
         "- If someone asks how the full bundle is generated, open examples/demos/06_booth_demo.py and then src/iints/analysis/booth_demo.py.\n"
         "  Reason: those files define the three scenarios and write the poster, talk track, and run bundle outputs.\n\n"

iints/analysis/clinical_metrics.py

@@ -144,6 +144,12 @@ class ClinicalMetricsCalculator:
         """
         self.target_range = target_range
         self.tight_range = tight_range
+
+    @staticmethod
+    def _percentage(mask: pd.Series) -> float:
+        if len(mask) == 0:
+            return 0.0
+        return float(mask.sum()) / float(len(mask)) * 100.0
         
     def calculate_tir(self, 
                       glucose: pd.Series, 
@@ -169,13 +175,13 @@ class ClinicalMetricsCalculator:
     def calculate_all_tir_metrics(self, glucose: pd.Series) -> Dict[str, float]:
         """Calculate all TIR-related metrics"""
         return {
-            'tir_70_180': self.calculate_tir(glucose, 70, 180),
-            'tir_70_140': self.calculate_tir(glucose, 70, 140),
-            'tir_70_110': self.calculate_tir(glucose, 70, 110),
-            'tir_below_70': self.calculate_tir(glucose, 0, 70),
-            'tir_below_54': self.calculate_tir(glucose, 0, 54),
-            'tir_above_180': self.calculate_tir(glucose, 180, 600),
-            'tir_above_250': self.calculate_tir(glucose, 250, 600)
+            'tir_70_180': self._percentage((glucose >= 70) & (glucose <= 180)),
+            'tir_70_140': self._percentage((glucose >= 70) & (glucose <= 140)),
+            'tir_70_110': self._percentage((glucose >= 70) & (glucose <= 110)),
+            'tir_below_70': self._percentage(glucose < 70),
+            'tir_below_54': self._percentage(glucose < 54),
+            'tir_above_180': self._percentage(glucose > 180),
+            'tir_above_250': self._percentage(glucose > 250)
         }
     
     def calculate_gmi(self, glucose: pd.Series) -> float:

iints/analysis/validator.py

@@ -4,6 +4,12 @@ from typing import Dict, List, Tuple, Optional, Any
 from dataclasses import dataclass
 from enum import Enum
 
+from iints.core.safety.config import (
+    SENSOR_GLUCOSE_MAX_MGDL,
+    SENSOR_GLUCOSE_MIN_MGDL,
+    SENSOR_MAX_GLUCOSE_RATE_PER_MIN_MGDL,
+)
+
 class ReliabilityLevel(Enum):
     HIGH = "high"
     MEDIUM = "medium" 
@@ -22,9 +28,9 @@ class DataIntegrityValidator:
     """Validates data integrity for reverse engineering analysis."""
     
     def __init__(self):
-        # Physiological limits
-        self.max_glucose_rate = 10  # mg/dL per minute
-        self.glucose_range = (20, 600)  # Physiologically possible range
+        # Broad CGM/sensor fail-soft limits
+        self.max_glucose_rate = SENSOR_MAX_GLUCOSE_RATE_PER_MIN_MGDL
+        self.glucose_range = (SENSOR_GLUCOSE_MIN_MGDL, SENSOR_GLUCOSE_MAX_MGDL)
         self.max_insulin_per_step = 5.0  # Units per 5-min step
         
     def validate_glucose_data(self, glucose_values: List[float], timestamps: List[float]) -> ValidationResult:
@@ -270,4 +276,4 @@ class ReverseEngineeringValidator:
             "issues": all_issues,
             "warnings": all_warnings,
             "category_scores": {category: result.reliability_score for category, result in validation_results.items()}
-        }
+        }

iints/cli/cli.py

@@ -53,6 +53,7 @@ from iints.data.registry import (
 )
 from iints.data.contracts import load_contract_yaml
 from iints.data.synthetic_mirror import generate_synthetic_mirror
+from iints.demo_assets import export_live_stage_demo
 from iints.mdmp.backend import (
     MDMP_GRADE_ORDER,
     active_mdmp_backend,
@@ -2740,6 +2741,39 @@ def demo_booth(
     )
 
 
+@app.command("demo-export")
+def demo_export(
+    output_dir: Annotated[
+        Path,
+        typer.Option(help="Directory where the bundled live stage demo files should be written."),
+    ] = Path("./iints_demo"),
+    overwrite: Annotated[
+        bool,
+        typer.Option("--overwrite/--no-overwrite", help="Allow overwriting exported demo files."),
+    ] = False,
+) -> None:
+    """Export the showable live demo code from the installed SDK."""
+    console = Console()
+    try:
+        outputs = export_live_stage_demo(output_dir=output_dir, overwrite=overwrite)
+    except FileExistsError as exc:
+        console.print(f"[bold red]Demo export stopped:[/bold red] {exc}")
+        raise typer.Exit(code=1)
+    except Exception as exc:
+        console.print(f"[bold red]Demo export failed:[/bold red] {exc}")
+        raise typer.Exit(code=1)
+
+    table = Table(title="IINTS Demo Export")
+    table.add_column("Artifact", style="cyan")
+    table.add_column("Path", overflow="fold")
+    table.add_row("script", outputs["script_path"])
+    table.add_row("notes", outputs["notes_path"])
+    console.print(table)
+    console.print(
+        "[green]Next:[/green] open `07_live_stage_demo.py`, explain the visible SDK calls, then run `python 07_live_stage_demo.py`."
+    )
+
+
 @app.command()
 def report(
     results_csv: Annotated[Path, typer.Option(help="Path to a simulation results CSV")],

iints/core/patient/bergman_model.py

@@ -46,7 +46,7 @@ class BergmanParameters:
     n: float = 0.23         # 1/min  — fractional insulin degradation
     Ib: float = 7.0         # mU/L   — basal plasma insulin
     Vi: float = 0.05        # L/kg   — insulin distribution volume
-    gamma: float = 0.004    # (mU/L)/(mg/dL)/min — endogenous secretion gain
+    gamma: float = 0.0      # (mU/L)/(mg/dL)/min — endogenous secretion gain (0 for T1D default)
     h: float = 80.0         # mg/dL  — secretion glucose threshold
 
     # --- Gut absorption ---
@@ -330,7 +330,8 @@ class BergmanPatientModel:
         dXdt = -p.p2 * X + p.p3 * max(I - p.Ib, 0.0)
 
         # --- dI/dt ---
-        # Endogenous pancreatic secretion (blunted in T1D, but kept for generality)
+        # T1D defaults keep endogenous secretion disabled (gamma=0). Override
+        # gamma explicitly if you intentionally want residual beta-cell function.
         secretion = p.gamma * max(G - p.h, 0.0)
         dIdt = -p.n * (I - p.Ib) + secretion + u_insulin_mu_per_min / Vi_abs
 

iints/core/patient/models.py

@@ -14,7 +14,7 @@ class CustomPatientModel:
     This model is intended for educational and stress-testing purposes, not for clinical accuracy.
     """
     def __init__(self, basal_insulin_rate: float = 0.8, insulin_sensitivity: float = 50.0,
-                 carb_factor: float = 10.0, glucose_decay_rate: float = 0.002,
+                 carb_factor: float = 10.0, glucose_decay_rate: float = 0.05,
                  initial_glucose: float = 120.0, glucose_absorption_rate: float = 0.03,
                  basal_glucose_target: Optional[float] = None,
                  insulin_action_duration: float = 300.0, # minutes, e.g., 5 hours

iints/core/patient/patient_factory.py

@@ -63,14 +63,14 @@ class PatientFactory:
         """Get a diverse set of patients for population studies."""
         if not SIMGLUCOSE_AVAILABLE:
             # Create diverse custom patients with different parameters
-            return [
-                CustomPatientModel(initial_glucose=120, insulin_sensitivity=40),  # High sensitivity
-                CustomPatientModel(initial_glucose=120, insulin_sensitivity=60),  # Low sensitivity
-                CustomPatientModel(initial_glucose=120, carb_factor=8),          # Fast carb absorption
-                CustomPatientModel(initial_glucose=120, carb_factor=12),         # Slow carb absorption
-                CustomPatientModel(initial_glucose=120, glucose_decay_rate=0.0015), # Slow metabolism
-                CustomPatientModel(initial_glucose=120, glucose_decay_rate=0.0035), # Fast metabolism
-            ]
+                return [
+                    CustomPatientModel(initial_glucose=120, insulin_sensitivity=40),  # High sensitivity
+                    CustomPatientModel(initial_glucose=120, insulin_sensitivity=60),  # Low sensitivity
+                    CustomPatientModel(initial_glucose=120, carb_factor=8),          # Fast carb absorption
+                    CustomPatientModel(initial_glucose=120, carb_factor=12),         # Slow carb absorption
+                    CustomPatientModel(initial_glucose=120, glucose_decay_rate=0.03), # Slower homeostatic drift
+                    CustomPatientModel(initial_glucose=120, glucose_decay_rate=0.07), # Faster homeostatic drift
+                ]
         else:
             # Use FDA-approved virtual patients
             selected_patients = [

iints/core/patient/profile.py

@@ -18,7 +18,7 @@ class PatientProfile:
     dawn_end_hour: float = 8.0
 
     # Advanced knobs (optional)
-    glucose_decay_rate: float = 0.002
+    glucose_decay_rate: float = 0.05
     glucose_absorption_rate: float = 0.03
     insulin_action_duration: float = 300.0
     insulin_peak_time: float = 75.0

iints/core/safety/config.py

@@ -2,16 +2,25 @@ from __future__ import annotations
 
 from dataclasses import dataclass
 
+SENSOR_GLUCOSE_MIN_MGDL = 40.0
+SENSOR_GLUCOSE_MAX_MGDL = 500.0
+SENSOR_MAX_GLUCOSE_DELTA_PER_5_MIN_MGDL = 20.0
+SENSOR_MAX_GLUCOSE_RATE_PER_MIN_MGDL = 4.0
+SIMULATION_GLUCOSE_FLOOR_MGDL = 20.0
+SIMULATION_GLUCOSE_CEILING_MGDL = 600.0
+
 
 @dataclass
 class SafetyConfig:
     """
     Central safety configuration for simulator, input validation, and supervisor.
     """
-    # Input validation limits
-    min_glucose: float = 20.0
-    max_glucose: float = 600.0
-    max_glucose_delta_per_5_min: float = 35.0
+    # Broad CGM/sensor plausibility limits.
+    # These are intentionally device-aware fail-soft bounds, not a claim about
+    # the full physiologic envelope of blood glucose inside the simulator.
+    min_glucose: float = SENSOR_GLUCOSE_MIN_MGDL
+    max_glucose: float = SENSOR_GLUCOSE_MAX_MGDL
+    max_glucose_delta_per_5_min: float = SENSOR_MAX_GLUCOSE_DELTA_PER_5_MIN_MGDL
 
     # Supervisor thresholds
     hypoglycemia_threshold: float = 70.0

iints/core/safety/input_validator.py

@@ -1,6 +1,11 @@
 from typing import Any, Dict, Optional
 
-from iints.core.safety.config import SafetyConfig
+from iints.core.safety.config import (
+    SENSOR_GLUCOSE_MAX_MGDL,
+    SENSOR_GLUCOSE_MIN_MGDL,
+    SENSOR_MAX_GLUCOSE_DELTA_PER_5_MIN_MGDL,
+    SafetyConfig,
+)
 
 class InputValidator:
     """
@@ -9,16 +14,16 @@ class InputValidator:
     This component makes the system robust against common sensor errors.
     """
     def __init__(self,
-                 min_glucose: float = 20.0,
-                 max_glucose: float = 600.0,
-                 max_glucose_delta_per_5_min: float = 35.0,
+                 min_glucose: float = SENSOR_GLUCOSE_MIN_MGDL,
+                 max_glucose: float = SENSOR_GLUCOSE_MAX_MGDL,
+                 max_glucose_delta_per_5_min: float = SENSOR_MAX_GLUCOSE_DELTA_PER_5_MIN_MGDL,
                  safety_config: Optional[SafetyConfig] = None):
         """
         Initializes the validator with plausible biological limits.
 
         Args:
-            min_glucose (float): The absolute minimum plausible glucose value (mg/dL).
-            max_glucose (float): The absolute maximum plausible glucose value (mg/dL).
+            min_glucose (float): Broad fail-soft lower bound for incoming CGM/sensor values (mg/dL).
+            max_glucose (float): Broad fail-soft upper bound for incoming CGM/sensor values (mg/dL).
             max_glucose_delta_per_5_min (float): The maximum plausible change in glucose
                                                  over a 5-minute period (mg/dL).
         """
@@ -62,7 +67,7 @@ class InputValidator:
         Raises:
             ValueError: If the value is outside biological plausibility limits.
         """
-        # 1. Absolute biological plausibility check
+        # 1. Broad CGM/sensor plausibility check
         if not (self.min_glucose <= glucose_value <= self.max_glucose):
             raise ValueError(
                 f"BIOLOGICAL_PLAUSIBILITY_ERROR: Glucose {glucose_value} mg/dL is outside the "

iints/data/adapter.py

@@ -10,6 +10,8 @@ import numpy as np
 from pathlib import Path
 from typing import Dict, List, Optional, Union, Any
 
+from iints.core.safety.config import SENSOR_GLUCOSE_MAX_MGDL, SENSOR_GLUCOSE_MIN_MGDL
+
 class DataAdapter:
     """Universal data adapter for IINTS-AF framework"""
     
@@ -51,8 +53,11 @@ class DataAdapter:
             raise ValueError(f"Missing required columns: {missing_cols}")
         
         # Validate glucose range
-        if df['glucose'].min() < 20 or df['glucose'].max() > 600:
-            raise ValueError("Glucose values outside physiological range (20-600 mg/dL)")
+        if df['glucose'].min() < SENSOR_GLUCOSE_MIN_MGDL or df['glucose'].max() > SENSOR_GLUCOSE_MAX_MGDL:
+            raise ValueError(
+                "Glucose values outside broad CGM/sensor-valid range "
+                f"({int(SENSOR_GLUCOSE_MIN_MGDL)}-{int(SENSOR_GLUCOSE_MAX_MGDL)} mg/dL)"
+            )
     
     def load_ohio_dataset(self, patient_id: str) -> pd.DataFrame:
         """Load Ohio T1DM dataset with clinical benchmarks"""

iints/data/ingestor.py

@@ -3,6 +3,8 @@ from pathlib import Path
 from typing import Dict, Any, Union, Optional
 import yaml
 
+from iints.core.safety.config import SENSOR_GLUCOSE_MAX_MGDL, SENSOR_GLUCOSE_MIN_MGDL
+
 class DataIngestor:
     """
     Standardized Data Bridge for ingesting various diabetes datasets into a
@@ -64,8 +66,11 @@ class DataIngestor:
         
         # Basic quality checks (from DATA_SCHEMA.md)
         if 'glucose' in df.columns:
-            if not ((df['glucose'] >= 20) & (df['glucose'] <= 600)).all():
-                raise ValueError("Glucose values outside acceptable range (20-600 mg/dL)")
+            if not ((df['glucose'] >= SENSOR_GLUCOSE_MIN_MGDL) & (df['glucose'] <= SENSOR_GLUCOSE_MAX_MGDL)).all():
+                raise ValueError(
+                    "Glucose values outside acceptable CGM/sensor range "
+                    f"({int(SENSOR_GLUCOSE_MIN_MGDL)}-{int(SENSOR_GLUCOSE_MAX_MGDL)} mg/dL)"
+                )
         if 'insulin' in df.columns and not df['insulin'].isna().all():
             if not ((df['insulin'] >= 0) & (df['insulin'] <= 50)).all():
                 raise ValueError("Insulin values outside acceptable range (0-50 units)")

iints/data/quality_checker.py

@@ -10,6 +10,12 @@ from datetime import datetime, timedelta
 import pandas as pd
 import numpy as np
 
+from iints.core.safety.config import (
+    SENSOR_GLUCOSE_MAX_MGDL,
+    SENSOR_GLUCOSE_MIN_MGDL,
+    SENSOR_MAX_GLUCOSE_RATE_PER_MIN_MGDL,
+)
+
 
 @dataclass
 class QualityReport:
@@ -100,14 +106,16 @@ class DataQualityChecker:
     
     # Physiological limits for glucose values
     GLUCOSE_LIMITS = {
-        'minimum': 20,    # mg/dL - physiologically possible minimum
-        'maximum': 600,   # mg/dL - physiologically possible maximum
+        'minimum': SENSOR_GLUCOSE_MIN_MGDL,  # mg/dL - broad CGM/sensor-valid minimum
+        'maximum': SENSOR_GLUCOSE_MAX_MGDL,  # mg/dL - broad CGM/sensor-valid maximum
         'critical_low': 54,   # mg/dL - clinically significant low
         'critical_high': 350  # mg/dL - clinically significant high
     }
-    
+
     PHYSIOLOGICAL_RATES = {
-        'max_glucose_change_per_min': 19.9 # mg/dL/min - Detecting changes of 20 mg/dL/min or more
+        # Broad fail-soft CGM plausibility cap; common CGM trend-arrow systems
+        # already treat >2 mg/dL/min as "rapid", so this keeps a conservative margin.
+        'max_glucose_change_per_min': SENSOR_MAX_GLUCOSE_RATE_PER_MIN_MGDL
     }
     
     # Expected sampling intervals (in minutes)

iints/data/registry.py

@@ -5,8 +5,10 @@ import urllib.request
 import zipfile
 import hashlib
 import shutil
+from ipaddress import ip_address
 from pathlib import Path
 from typing import Any, Dict, List, Optional, IO, cast
+from urllib.parse import urlparse
 
 try:  # Python 3.9+
     from importlib.resources import files
@@ -54,10 +56,30 @@ def list_dataset_ids() -> List[str]:
     return ids
 
 
+def _is_loopback_host(hostname: str) -> bool:
+    if hostname == "localhost":
+        return True
+    try:
+        return ip_address(hostname).is_loopback
+    except ValueError:
+        return False
+
+
+def _validate_download_url(url: str) -> str:
+    parsed = urlparse(url)
+    if parsed.scheme == "https":
+        return url
+    if parsed.scheme == "http" and parsed.hostname and _is_loopback_host(parsed.hostname):
+        return url
+    raise DatasetFetchError(
+        "Dataset download URL must use https, or http only for localhost/loopback development mirrors."
+    )
+
+
 def _download_file(url: str, output_path: Path) -> Path:
     output_path.parent.mkdir(parents=True, exist_ok=True)
     try:
-        urllib.request.urlretrieve(url, output_path)
+        urllib.request.urlretrieve(_validate_download_url(url), output_path)  # nosec B310 - URL is scheme validated before download
     except Exception as exc:
         raise DatasetFetchError(f"Failed to download {url}: {exc}") from exc
     return output_path

iints/demo_assets.py

@@ -0,0 +1,50 @@
+from __future__ import annotations
+
+from importlib.resources import files
+from pathlib import Path
+
+
+def export_live_stage_demo(output_dir: str | Path = '.', *, overwrite: bool = False) -> dict[str, str]:
+    """Export the bundled live stage demo script for users running the installed SDK."""
+    resolved_output = Path(output_dir).expanduser().resolve()
+    resolved_output.mkdir(parents=True, exist_ok=True)
+
+    script_target = resolved_output / '07_live_stage_demo.py'
+    notes_target = resolved_output / 'RUN_ME_FIRST.txt'
+
+    if not overwrite and script_target.exists():
+        raise FileExistsError(f'Demo script already exists: {script_target}')
+    if not overwrite and notes_target.exists():
+        raise FileExistsError(f'Instruction file already exists: {notes_target}')
+
+    script_content = files('iints.templates.demos').joinpath('live_stage_demo.py').read_text(encoding='utf-8')
+    notes_content = (
+        'IINTS LIVE DEMO EXPORT\n'
+        '======================\n\n'
+        'This folder was exported from the installed IINTS SDK.\n\n'
+        '1. Activate the virtual environment that contains IINTS.\n'
+        '2. Open 07_live_stage_demo.py and point to:\n'
+        '   - PATIENT_CONFIG\n'
+        '   - OUTPUT_DIR\n'
+        '   - DURATION_MINUTES\n'
+        '   - TIME_STEP_MINUTES\n'
+        '   - SEED\n'
+        '3. Explain that the script visibly calls:\n'
+        '   - run_full(...)\n'
+        '   - generate_results_poster(...)\n'
+        '   - prepare_ai_ready_artifacts(...)\n'
+        '4. Run:\n'
+        '   python 07_live_stage_demo.py\n'
+        '5. Open the generated files under results/booth_demo_live/.\n\n'
+        'Tip: if you also cloned the SDK repo, you can run the repo wrapper instead:\n'
+        '  ./scripts/run_live_stage_demo.sh\n'
+    )
+
+    script_target.write_text(script_content, encoding='utf-8')
+    notes_target.write_text(notes_content, encoding='utf-8')
+
+    return {
+        'output_dir': str(resolved_output),
+        'script_path': str(script_target),
+        'notes_path': str(notes_target),
+    }

iints/highlevel.py

@@ -442,7 +442,7 @@ def run_population(
         dawn_phenomenon_strength=patient_params.get("dawn_phenomenon_strength", 0.0),
         dawn_start_hour=patient_params.get("dawn_start_hour", 4.0),
         dawn_end_hour=patient_params.get("dawn_end_hour", 8.0),
-        glucose_decay_rate=patient_params.get("glucose_decay_rate", 0.002),
+        glucose_decay_rate=patient_params.get("glucose_decay_rate", 0.05),
         glucose_absorption_rate=patient_params.get("glucose_absorption_rate", 0.03),
         insulin_peak_time=patient_params.get("insulin_peak_time", 75.0),
         meal_mismatch_epsilon=patient_params.get("meal_mismatch_epsilon", 1.0),

iints/learning/autonomous_optimizer.py

@@ -15,6 +15,18 @@ except Exception:  # pragma: no cover
     _TORCH_AVAILABLE = False
 
 
+def _safe_torch_load_weights(path: str):
+    if torch is None:  # pragma: no cover
+        raise ImportError("Torch required for model loading.")
+    try:
+        return torch.load(path, map_location="cpu", weights_only=True)
+    except TypeError as exc:
+        raise RuntimeError(
+            "This PyTorch build does not support secure weights-only loading. "
+            "Upgrade torch to a version that supports `weights_only=True`."
+        ) from exc
+
+
 @dataclass
 class ClinicalConstraints:
     """Physiological constraints based on medical literature."""
@@ -98,7 +110,7 @@ if _TORCH_AVAILABLE:
             model = LSTMModel(input_size=7, hidden_size=50, output_size=1)
 
             if os.path.exists(self.model_path):
-                model.load_state_dict(torch.load(self.model_path))
+                model.load_state_dict(_safe_torch_load_weights(self.model_path))
 
             improved_model = self._clinical_fine_tuning(model, clinical_X, clinical_y)
             safety_score = self._validate_clinical_safety(improved_model, clinical_X, clinical_y)

iints/mdmp/backend.py

@@ -114,9 +114,8 @@ def get_backend() -> str:
     if requested in {BACKEND_MDMP, "mdmp", "external"}:
         if not is_mdmp_available():
             raise ImportError(
-                "mdmp_core not found.\n"
-                "Install with: pip install iints-sdk-python35[mdmp]\n"
-                "or: pip install 'mdmp-protocol>=0.3.0'"
+                "Bundled MDMP support is not available in this environment.\n"
+                "Install with: pip install iints-sdk-python35[mdmp]"
             )
         return BACKEND_MDMP
     if requested == "auto" and is_mdmp_available():

iints/research/predictor.py

@@ -293,7 +293,15 @@ def load_predictor(model_path: Path) -> Tuple["LSTMPredictor", dict]:
         raise ImportError(
             "Torch is required for predictor loading. Install with `pip install iints-sdk-python35[research]`."
         ) from _IMPORT_ERROR
-    payload = torch.load(model_path, map_location="cpu", weights_only=False)
+    try:
+        payload = torch.load(model_path, map_location="cpu", weights_only=True)
+    except TypeError as exc:
+        raise RuntimeError(
+            "This PyTorch build does not support secure checkpoint loading. "
+            "Upgrade torch to a version with `weights_only=True` support."
+        ) from exc
+    if not isinstance(payload, dict):
+        raise RuntimeError("Predictor checkpoint must decode to a dictionary payload.")
     config = payload["config"]
     model = LSTMPredictor(
         input_size=config["input_size"],

iints/templates/demos/__init__.py

@@ -0,0 +1 @@
+"""Bundled demo script templates for installed IINTS users."""