Nigeria Health Pack

Overview

NigeriaHealthcareCompliance enforces Nigerian healthcare AI regulations simultaneously, including patient consent, clinical oversight requirements, and special-category health data restrictions under the NDPA.

from comply54.sectors import NigeriaHealthcareCompliance

compliance = NigeriaHealthcareCompliance()
result = compliance.check(
    action="generate_diagnosis",
    params={"patient_id": "P001", "symptoms": ["fever", "headache"]},
    context={"human_clinician_oversight": True, "consent_documented": True},
)

Included packs

Pack ID	Regulation	Authority	Key controls
`nigeria/nha`	National Health Act 2014	FMOH	Patient consent (§26), confidentiality (§29), prescription prohibition (§30), cross-border health data
`nigeria/ndpa`	NDPA 2023 — special-category	NITDA	Health data as special-category, cross-border restrictions (§25), consent
`nigeria/bvn-nin`	BVN/NIN Framework	CBN/NIMC	Biometric PII detection in outputs
`universal/pii-leakage`	OWASP LLM01	OWASP	Patient ID, health record numbers, diagnoses in outputs
`universal/prompt-injection`	OWASP LLM01	OWASP	Jailbreak and indirect injection patterns
`universal/tool-permissions`	OWASP LLM08	OWASP	Excessive tool scope, unauthorised record access
`universal/human-approval`	OWASP LLM08	OWASP	Irreversible clinical actions require human sign-off

Decision matrix by action type

Patient data access

Condition	Decision	Reason
`consent_documented = true`	`allow` / further checks	NHA §26 consent present
`consent_documented = false`	`deny`	NHA §26: patient consent required before accessing health records
Cross-border, health data	`deny`	NDPA §25 + NHA §29: health data is special-category
Bulk access > 10 records	`escalate`	NHA §29: bulk health record access requires escalation

AI diagnosis

Condition	Decision	Reason
`human_clinician_oversight = true`	`escalate`	FMOH AI Policy Guideline 4: clinician sign-off required
`human_clinician_oversight = false`	`deny`	FMOH Guideline 4: AI diagnosis without clinician oversight prohibited

Prescriptions

Condition	Decision	Reason
`licensed_clinician_approval = true`	`allow`	MDP Act Cap M8 §16 met
`licensed_clinician_approval = false`	`deny`	NHA §30 + MDP Act §16: only licensed clinicians may prescribe

Research / analytics

Condition	Decision	Reason
Purpose = `research`	`escalate`	NHA §29: IRB/ethics approval required for research use
Purpose = `treatment`	Standard checks	Normal clinical workflow

Usage

Standalone check

from comply54.sectors import NigeriaHealthcareCompliance

compliance = NigeriaHealthcareCompliance()

# Access patient records — consent required
result = compliance.check(
    action="access_patient_records",
    params={"patient_id": "P001", "record_type": "full_history"},
    context={"consent_documented": True},
)

if result.blocked:
    print(result.primary_violation.messages[0])
elif result.overall == "escalate":
    print(f"Escalation required — audit ID: {result.audit_id}")

With LangGraph (ReAct agent)

from comply54.sectors import NigeriaHealthcareCompliance
from comply54.langchain import Comply54Guard, comply54_route
from langgraph.graph import StateGraph, MessagesState
from langgraph.prebuilt import ToolNode

compliance = NigeriaHealthcareCompliance()
tools = [access_patient_records_tool, generate_diagnosis_tool, prescribe_medication_tool]
guard = Comply54Guard(compliance, tools)

def build_graph():
    graph = StateGraph(MessagesState)
    graph.add_node("agent", agent_node)
    graph.add_node("guard", guard)
    graph.add_node("tools", ToolNode(tools))

    graph.set_entry_point("agent")
    graph.add_conditional_edges("agent", comply54_route, {"guard": "guard", "__end__": "__end__"})
    graph.add_edge("guard", "tools")
    graph.add_edge("tools", "agent")
    return graph.compile()

# Pass clinical context via compliance_context state key
result = build_graph().invoke({
    "messages": [("user", "Generate a diagnosis for patient P001")],
    "compliance_context": {
        "consent_documented": True,
        "human_clinician_oversight": False,  # will be denied by FMOH Guideline 4
    },
})

Strict mode

In strict mode, escalate decisions are upgraded to deny — no clinical action proceeds without full human approval:

compliance = NigeriaHealthcareCompliance(strict_mode=True)

Accessing all decisions

result = compliance.check(
    action="generate_diagnosis",
    params={"patient_id": "P001"},
    context={"human_clinician_oversight": True, "consent_documented": True},
)

for decision in result.decisions:
    status = "✓" if decision.action == "allow" else "!"
    print(f"{status} {decision.pack}: {decision.action}")
    for msg in decision.messages:
        print(f"   {msg}")

Properties

compliance = NigeriaHealthcareCompliance()
compliance.name           # "Nigeria Healthcare Compliance"
compliance.jurisdictions  # ["NG"]
len(compliance.packs)     # 7

Overview​

Included packs​

Decision matrix by action type​

Patient data access​

AI diagnosis​

Prescriptions​

Research / analytics​

Usage​

Standalone check​

With LangGraph (ReAct agent)​

Strict mode​

Accessing all decisions​

Properties​

Related​