empathy-framework 3.7.0__py3-none-any.whl → 3.8.0__py3-none-any.whl

empathy_healthcare_plugin/monitors/monitoring/trajectory_analyzer.py

@@ -0,0 +1,389 @@
+"""Trajectory Analyzer (Level 4)
+
+Analyzes vital sign trends to predict patient deterioration BEFORE critical.
+
+This is Level 4 Anticipatory Empathy - alerting before the patient meets full crisis criteria.
+
+Copyright 2025 Smart AI Memory, LLC
+Licensed under Fair Source 0.9
+"""
+
+from dataclasses import dataclass
+from typing import Any
+
+
+@dataclass
+class VitalTrend:
+    """Trend analysis for a single vital sign"""
+
+    parameter: str
+    current_value: float
+    previous_value: float
+    change: float
+    change_percent: float
+    direction: str  # "increasing", "decreasing", "stable"
+    rate_of_change: float  # units per hour
+    concerning: bool
+    reasoning: str
+
+
+@dataclass
+class TrajectoryPrediction:
+    """Prediction of patient trajectory.
+
+    This is Level 4 - predicting BEFORE criteria met.
+    """
+
+    trajectory_state: str  # "stable", "improving", "concerning", "critical"
+    estimated_time_to_critical: str | None
+    vital_trends: list[VitalTrend]
+    overall_assessment: str
+    confidence: float
+    recommendations: list[str]
+
+
+class TrajectoryAnalyzer:
+    """Analyzes vital sign trajectory to predict deterioration.
+
+    This implements Level 4 Anticipatory Empathy.
+    """
+
+    def __init__(self):
+        # Define normal ranges
+        self.normal_ranges = {
+            "hr": (60, 100),
+            "systolic_bp": (90, 140),
+            "diastolic_bp": (60, 90),
+            "respiratory_rate": (12, 20),
+            "temp_f": (97.0, 99.5),
+            "o2_sat": (95, 100),
+        }
+
+        # Define concerning rates of change
+        self.concerning_rates = {
+            "hr": 15,  # bpm increase over 2 hours
+            "systolic_bp": 20,  # mmHg decrease
+            "respiratory_rate": 5,  # breaths/min increase
+            "temp_f": 2.0,  # degrees increase
+        }
+
+    def analyze_trajectory(
+        self,
+        current_data: dict[str, float],
+        historical_data: list[dict[str, Any]],
+    ) -> TrajectoryPrediction:
+        """Analyze patient trajectory from historical vitals.
+
+        Args:
+            current_data: Current vital signs
+            historical_data: List of previous readings (last 6-12 hours)
+
+        Returns:
+            TrajectoryPrediction with assessment
+
+        Example:
+            >>> history = [
+            ...     {"timestamp": "12:00", "hr": 95, "systolic_bp": 120},
+            ...     {"timestamp": "13:00", "hr": 105, "systolic_bp": 110},
+            ...     {"timestamp": "14:00", "hr": 112, "systolic_bp": 95}
+            ... ]
+            >>> prediction = analyzer.analyze_trajectory(current_vitals, history)
+            >>> if prediction.trajectory_state == "concerning":
+            ...     print(f"ALERT: {prediction.overall_assessment}")
+
+        """
+        if not historical_data:
+            return TrajectoryPrediction(
+                trajectory_state="stable",
+                estimated_time_to_critical=None,
+                vital_trends=[],
+                overall_assessment="Insufficient historical data for trajectory analysis",
+                confidence=0.3,
+                recommendations=["Continue monitoring"],
+            )
+
+        # Analyze trends for each vital sign
+        vital_trends = []
+
+        for parameter, current_value in current_data.items():
+            if parameter in ["mental_status"]:  # Skip non-numeric
+                continue
+
+            trend = self._analyze_parameter_trend(parameter, current_value, historical_data)
+
+            if trend:
+                vital_trends.append(trend)
+
+        # Determine overall trajectory state
+        trajectory_state = self._determine_trajectory_state(vital_trends)
+
+        # Estimate time to critical (if concerning)
+        time_to_critical = None
+        if trajectory_state in ["concerning", "critical"]:
+            time_to_critical = self._estimate_time_to_critical(vital_trends, current_data)
+
+        # Generate overall assessment
+        assessment = self._generate_assessment(trajectory_state, vital_trends, time_to_critical)
+
+        # Generate recommendations
+        recommendations = self._generate_recommendations(trajectory_state, vital_trends)
+
+        # Calculate confidence
+        confidence = self._calculate_confidence(historical_data, vital_trends)
+
+        return TrajectoryPrediction(
+            trajectory_state=trajectory_state,
+            estimated_time_to_critical=time_to_critical,
+            vital_trends=vital_trends,
+            overall_assessment=assessment,
+            confidence=confidence,
+            recommendations=recommendations,
+        )
+
+    def _analyze_parameter_trend(
+        self,
+        parameter: str,
+        current_value: float,
+        historical_data: list[dict[str, Any]],
+    ) -> VitalTrend | None:
+        """Analyze trend for single parameter"""
+        # Extract historical values
+        historical_values = []
+        for entry in historical_data:
+            if parameter in entry and entry[parameter] is not None:
+                historical_values.append(entry[parameter])
+
+        if not historical_values:
+            return None
+
+        # Calculate change from most recent
+        previous_value = historical_values[-1]
+        change = current_value - previous_value
+        change_percent = (change / previous_value * 100) if previous_value != 0 else 0
+
+        # Determine direction
+        if abs(change_percent) < 5:
+            direction = "stable"
+        elif change > 0:
+            direction = "increasing"
+        else:
+            direction = "decreasing"
+
+        # Calculate rate of change (per hour)
+        # Assuming historical data spans 2-6 hours
+        hours_elapsed = len(historical_values) / 2  # Rough estimate
+        rate_of_change = abs(change) / hours_elapsed if hours_elapsed > 0 else 0
+
+        # Determine if concerning
+        concerning, reasoning = self._is_trend_concerning(
+            parameter,
+            current_value,
+            change,
+            rate_of_change,
+            direction,
+        )
+
+        return VitalTrend(
+            parameter=parameter,
+            current_value=current_value,
+            previous_value=previous_value,
+            change=change,
+            change_percent=change_percent,
+            direction=direction,
+            rate_of_change=rate_of_change,
+            concerning=concerning,
+            reasoning=reasoning,
+        )
+
+    def _is_trend_concerning(
+        self,
+        parameter: str,
+        current_value: float,
+        change: float,
+        rate_of_change: float,
+        direction: str,
+    ) -> tuple[bool, str]:
+        """Determine if trend is concerning"""
+        # Check if currently out of normal range
+        if parameter in self.normal_ranges:
+            min_val, max_val = self.normal_ranges[parameter]
+
+            if current_value < min_val:
+                return True, f"{parameter} below normal range ({min_val}-{max_val})"
+            if current_value > max_val:
+                return True, f"{parameter} above normal range ({min_val}-{max_val})"
+
+        # Check rate of change
+        if parameter in self.concerning_rates:
+            threshold = self.concerning_rates[parameter]
+
+            if parameter == "hr" and direction == "increasing" and rate_of_change > threshold:
+                return True, f"HR increasing rapidly (+{change:.0f} bpm)"
+
+            if (
+                parameter == "systolic_bp"
+                and direction == "decreasing"
+                and rate_of_change > threshold
+            ):
+                return True, f"BP decreasing rapidly (-{abs(change):.0f} mmHg)"
+
+            if (
+                parameter == "respiratory_rate"
+                and direction == "increasing"
+                and rate_of_change > threshold
+            ):
+                return True, f"RR increasing rapidly (+{change:.0f} /min)"
+
+            if parameter == "temp_f" and direction == "increasing" and rate_of_change > threshold:
+                return True, f"Temp increasing rapidly (+{change:.1f}°F)"
+
+        return False, "Within normal trajectory"
+
+    def _determine_trajectory_state(self, vital_trends: list[VitalTrend]) -> str:
+        """Determine overall trajectory state"""
+        concerning_trends = [t for t in vital_trends if t.concerning]
+
+        if not concerning_trends:
+            return "stable"
+
+        # Count concerning trends by severity
+        critical_parameters = ["systolic_bp", "o2_sat"]
+        critical_concerning = sum(
+            1 for t in concerning_trends if t.parameter in critical_parameters
+        )
+
+        if critical_concerning >= 1:
+            return "critical"
+
+        if len(concerning_trends) >= 2:
+            return "concerning"
+
+        if len(concerning_trends) == 1:
+            return "concerning"
+
+        return "stable"
+
+    def _estimate_time_to_critical(
+        self,
+        vital_trends: list[VitalTrend],
+        current_data: dict[str, float],
+    ) -> str | None:
+        """Estimate time until patient meets critical criteria.
+
+        This is core Level 4 - predicting the future.
+        """
+        # Example: If BP dropping at 10 mmHg/hour, currently 95, critical is 85
+        # Time to critical = (95 - 85) / 10 = 1 hour
+
+        for trend in vital_trends:
+            if not trend.concerning:
+                continue
+
+            if trend.parameter == "systolic_bp" and trend.direction == "decreasing":
+                critical_threshold = 90
+                current = trend.current_value
+                rate = trend.rate_of_change
+
+                if rate > 0:
+                    hours_to_critical = (current - critical_threshold) / rate
+                    if 0 < hours_to_critical < 24:
+                        return f"~{int(hours_to_critical)} hours"
+
+            if trend.parameter == "o2_sat" and trend.direction == "decreasing":
+                critical_threshold = 90
+                current = trend.current_value
+                rate = trend.rate_of_change
+
+                if rate > 0:
+                    hours_to_critical = (current - critical_threshold) / rate
+                    if 0 < hours_to_critical < 24:
+                        return f"~{int(hours_to_critical)} hours"
+
+        return None
+
+    def _generate_assessment(
+        self,
+        trajectory_state: str,
+        vital_trends: list[VitalTrend],
+        time_to_critical: str | None,
+    ) -> str:
+        """Generate overall assessment"""
+        if trajectory_state == "stable":
+            return "Patient vitals stable. Continue routine monitoring."
+
+        concerning = [t for t in vital_trends if t.concerning]
+
+        if trajectory_state == "critical":
+            trends_desc = ", ".join(f"{t.parameter} {t.direction}" for t in concerning[:3])
+            return (
+                f"CRITICAL trajectory detected: {trends_desc}. Immediate intervention recommended."
+            )
+
+        if trajectory_state == "concerning":
+            trends_desc = ", ".join(f"{t.parameter} {t.direction}" for t in concerning[:3])
+
+            if time_to_critical:
+                return (
+                    f"Concerning trajectory: {trends_desc}. "
+                    f"In our experience, this pattern suggests deterioration. "
+                    f"Estimated time to critical: {time_to_critical}. "
+                    "Early intervention may prevent escalation."
+                )
+
+            return (
+                f"Concerning trajectory: {trends_desc}. "
+                "In our experience, this pattern warrants closer monitoring."
+            )
+
+        return "Patient trajectory under assessment."
+
+    def _generate_recommendations(
+        self,
+        trajectory_state: str,
+        vital_trends: list[VitalTrend],
+    ) -> list[str]:
+        """Generate actionable recommendations"""
+        if trajectory_state == "stable":
+            return ["Continue routine monitoring"]
+
+        recommendations = []
+
+        if trajectory_state in ["concerning", "critical"]:
+            recommendations.append("Notify physician of trajectory")
+            recommendations.append("Increase monitoring frequency")
+
+        concerning = [t for t in vital_trends if t.concerning]
+
+        for trend in concerning:
+            if trend.parameter == "systolic_bp":
+                recommendations.append("Assess volume status and perfusion")
+            elif trend.parameter == "hr":
+                recommendations.append("Assess for infection or pain")
+            elif trend.parameter == "respiratory_rate":
+                recommendations.append("Assess respiratory status and oxygenation")
+            elif trend.parameter == "temp_f":
+                recommendations.append("Assess for infection")
+
+        if trajectory_state == "critical":
+            recommendations.append("Consider rapid response team activation")
+
+        return recommendations
+
+    def _calculate_confidence(
+        self,
+        historical_data: list[dict[str, Any]],
+        vital_trends: list[VitalTrend],
+    ) -> float:
+        """Calculate confidence in prediction"""
+        # More data = higher confidence
+        data_points = len(historical_data)
+        data_confidence = min(data_points / 10, 1.0)
+
+        # More consistent trends = higher confidence
+        if vital_trends:
+            concerning_count = sum(1 for t in vital_trends if t.concerning)
+            trend_confidence = concerning_count / len(vital_trends)
+        else:
+            trend_confidence = 0.5
+
+        return (data_confidence + trend_confidence) / 2

empathy_llm_toolkit/agent_factory/__init__.py

@@ -0,0 +1,53 @@
+"""Empathy Framework - Universal Agent Factory
+
+Create agents using your preferred framework while retaining Empathy's
+cost optimization, pattern learning, and memory features.
+
+Supported Frameworks:
+- LangChain: Chains, tools, and retrieval
+- LangGraph: Stateful multi-agent graphs
+- AutoGen: Conversational multi-agent systems
+- Haystack: RAG and document pipelines
+- Native: Empathy's built-in agent system
+
+Usage:
+    from empathy_llm_toolkit.agent_factory import AgentFactory, Framework
+
+    # Create factory with preferred framework
+    factory = AgentFactory(framework=Framework.LANGGRAPH)
+
+    # Create agents
+    researcher = factory.create_agent("researcher", tools=[...])
+    writer = factory.create_agent("writer", model_tier="premium")
+
+    # Create workflows
+    pipeline = factory.create_workflow([researcher, writer])
+
+    # Create wizards with framework backing
+    debug_wizard = factory.create_wizard("debugging")
+
+Copyright 2025 Smart-AI-Memory
+Licensed under Fair Source License 0.9
+"""
+
+from empathy_llm_toolkit.agent_factory.base import (
+                                                    AgentCapability,
+                                                    AgentConfig,
+                                                    AgentRole,
+                                                    BaseAdapter,
+                                                    BaseAgent,
+                                                    WorkflowConfig,
+)
+from empathy_llm_toolkit.agent_factory.factory import AgentFactory
+from empathy_llm_toolkit.agent_factory.framework import Framework
+
+__all__ = [
+    "AgentCapability",
+    "AgentConfig",
+    "AgentFactory",
+    "AgentRole",
+    "BaseAdapter",
+    "BaseAgent",
+    "Framework",
+    "WorkflowConfig",
+]

empathy_llm_toolkit/agent_factory/adapters/__init__.py

@@ -0,0 +1,85 @@
+"""Framework Adapters for Agent Factory
+
+Each adapter implements the BaseAdapter interface for a specific
+agent framework, allowing seamless switching between frameworks.
+
+Copyright 2025 Smart-AI-Memory
+Licensed under Fair Source License 0.9
+"""
+
+from empathy_llm_toolkit.agent_factory.adapters.native import NativeAdapter
+from empathy_llm_toolkit.agent_factory.adapters.wizard_adapter import (
+    WizardAdapter,
+    WizardAgent,
+    wrap_wizard,
+)
+
+# Lazy imports for optional frameworks
+_langchain_adapter = None
+_langgraph_adapter = None
+_autogen_adapter = None
+_haystack_adapter = None
+_crewai_adapter = None
+
+
+def get_langchain_adapter():
+    """Get LangChain adapter (lazy import)."""
+    global _langchain_adapter
+    if _langchain_adapter is None:
+        from empathy_llm_toolkit.agent_factory.adapters.langchain_adapter import LangChainAdapter
+
+        _langchain_adapter = LangChainAdapter
+    return _langchain_adapter
+
+
+def get_langgraph_adapter():
+    """Get LangGraph adapter (lazy import)."""
+    global _langgraph_adapter
+    if _langgraph_adapter is None:
+        from empathy_llm_toolkit.agent_factory.adapters.langgraph_adapter import LangGraphAdapter
+
+        _langgraph_adapter = LangGraphAdapter
+    return _langgraph_adapter
+
+
+def get_autogen_adapter():
+    """Get AutoGen adapter (lazy import)."""
+    global _autogen_adapter
+    if _autogen_adapter is None:
+        from empathy_llm_toolkit.agent_factory.adapters.autogen_adapter import AutoGenAdapter
+
+        _autogen_adapter = AutoGenAdapter
+    return _autogen_adapter
+
+
+def get_haystack_adapter():
+    """Get Haystack adapter (lazy import)."""
+    global _haystack_adapter
+    if _haystack_adapter is None:
+        from empathy_llm_toolkit.agent_factory.adapters.haystack_adapter import HaystackAdapter
+
+        _haystack_adapter = HaystackAdapter
+    return _haystack_adapter
+
+
+def get_crewai_adapter():
+    """Get CrewAI adapter (lazy import)."""
+    global _crewai_adapter
+    if _crewai_adapter is None:
+        from empathy_llm_toolkit.agent_factory.adapters.crewai_adapter import CrewAIAdapter
+
+        _crewai_adapter = CrewAIAdapter
+    return _crewai_adapter
+
+
+__all__ = [
+    "NativeAdapter",
+    "WizardAdapter",
+    "WizardAgent",
+    "get_autogen_adapter",
+    "get_crewai_adapter",
+    "get_haystack_adapter",
+    "get_langchain_adapter",
+    "get_langgraph_adapter",
+    "wrap_wizard",
+]