Introduction
In most hospitals around the world, beds are simultaneously the most critical and the most poorly managed resource. Patient flow stalls not because of clinical complexity alone, but because of the operational latency embedded in how beds are tracked, allocated, cleaned, and communicated. The consequences ripple across every metric that matters to healthcare leaders: average length of stay, emergency department wait times, inefficient patient flows, hospital-acquired infection rates, operating costs, and staff morale.
Acuvate developed the Healthcare Bed Management Framework to address this challenge not with incremental process improvement, but with a fundamentally reimagined approach built on Agentic AI, real-time IoT sensing, and the enterprise-grade intelligence capabilities of Microsoft Fabric. This framework is not a prototype. It has been implemented across multiple healthcare organisations and is delivering measurable operational improvements at scale today.
The Acuvate Healthcare Bed Management Framework is a production-deployed, AI-native platform that has transformed bed operations for healthcare providers reducing length of stay, eliminating manual coordination overhead, and delivering real-time operational visibility across the entire hospital estate.
The Problem: Bed Inefficiency Is a Strategic Risk
Hospital bed management sits at the intersection of clinical care, operational logistics, and financial performance. When it functions well, patients flow smoothly from admission to discharge, beds turn over efficiently, Length of Stay (LOS) reduces, discharge planning improves, more inpatient bed capacity is created so it could effectively treat more patients and staff focus on care rather than coordination. When it fails as it routinely does the consequences are immediate and far-reaching.
Consider the typical discharge process: a patient is clinically ready to leave, but the bed remains occupied for two to four hours while housekeeping is manually notified, a clean is completed, and the bed is updated in a system that frontline staff have learned not to trust. In a 400-bed hospital, this delay replicated dozens of times per day translates to thousands of lost bed-hours every week. For patients waiting in the Emergency Department, those hours mean deteriorating clinical conditions and avoidable diversions.
The Operational Cost Reality
Industry benchmarks indicate hospitals lose between $200 and $300 per hour for every staffed bed that sits unnecessarily vacant. Suboptimal bed management contributes directly to extended average length of stay, elevated hospital-acquired infection risk from inappropriate placements, increased staff overtime, and declining patient satisfaction scores all of which carry direct financial penalties under value-based care models.
Why Traditional Systems Keep Failing
The healthcare industry has invested heavily in bed management tooling over the past two decades. Most major EHR platforms include a capacity management module. Dedicated bed boards and digital whiteboards are commonplace. Yet the problem persists because every traditional system shares the same fundamental design constraint: they present information to humans and wait for humans to act.
They are passive reporting tools designed around a human-in-the-loop workflow that was never going to eliminate coordination latency. The sequence clinician marks a discharge, coordinator notices, a call is placed to housekeeping, a cleaner arrives, someone updates the system remains unchanged regardless of how sophisticated the dashboard that surfaces it may be.
Traditional tool does not optimally utilize assets, and decisions are hard to make in real-time. Staff lacks the resources they needed to properly care for patients, which further contributes to burnout. In addition, delays in care progression detracted from the overall experience.
Furthermore, traditional tools cannot perform the multi-variable reasoning that truly optimal bed placement demands. Matching a patient’s acuity level, infection control requirements, proximity to specialist staff, required monitoring equipment, and anticipated length of stay simultaneously, across an entire hospital, in real time exceeds what any coordinator can consistently achieve under operational pressure. The result is reactive, suboptimal placement that propagates inefficiency through the entire care pathway.
The Framework: Agentic AI That Acts, Not Advises
Agentic AI represents a qualitative leap beyond conventional automation. Where traditional tools provide dashboards and alerts for humans to act upon, Agentic AI takes goal-directed action autonomously reasoning across dynamic context, adapting to changing conditions, and executing decisions within defined boundaries without waiting for human initiation.
The Acuvate Healthcare Bed Management Framework deploys two purpose-built autonomous agents through Microsoft Copilot Studio, orchestrated within an Automated Patient Admittance Agent architecture. Together, they manage the complete bed allocation lifecycle This Agentic Solution is deployed in Teams where staff from departments can chat.
Agent 1: Bed Recommendation
- Receives the full admission request patient age, diagnosis, acuity, isolation requirements, and care pathway either via the admin chatting with agentic ai bot deployed in teams or the agent pulls from the HER when a new admission is entered in it.
- Reasons across all available beds, evaluating clinical fit, ward suitability, equipment availability, and infection control constraints.
- Recommends the optimal bed not simply the first available based on holistic clinical and operational context. Passes the recommendation to Agent 2.
Agent 2: Bed Readiness & Placement
- Validates the recommended bed’s current housekeeping and readiness status using real time data. Get the Approval of Admin.
- Autonomously reserves the bed in the system, eliminating the manual update step that creates coordination latency today.
- Dispatches role-specific notifications to nursing, housekeeping, and transport via Microsoft Teams and Outlook.
The Critical Distinction
Traditional automation executes predefined rules. Agentic AI reasons across real-world context and takes goal-directed action. The difference is between a rigid decision tree and a capable, always-on operational colleague one that never goes off shift, never misses a status update, and processes the full complexity of the hospital floor simultaneously.
Real-Time IoT Intelligence: Ground Truth on Every Bed
One of the most persistent failure modes in bed management is the gap between what the system records and what is physically happening on the ward floor. A bed is marked as clean before housekeeping has finished. A patient is recorded as discharged but is still physically present. A bed is shown as available while a family member is still in the room collecting belongings. These discrepancies individually minor, collectively systemic cause coordinators to lose confidence in digital systems and revert to phone-based verification. The Acuvate framework resolves this through a dual-sensor IoT layer that provides continuous, objective occupancy data independent of any manual update or EHR entry.
Pressure Mat Sensor
- Embedded in each smart hospital bed. Provides a continuous, real-time occupancy signal eliminating ambiguity about whether a bed is physically occupied.
- Triggers automatic status transitions:
vacant → occupied on patient placement;
occupied → pending clean on departure.
RFID Patient Wristband
- Worn by each patient from admission. Confirms patient identity and physical location throughout their stay in real time.
- Enables accurate admission-to-discharge timeline capture for LOS analysis and trigger bed vacant alerts.
Sensor telemetry is transmitted via an ESP32 Wi-Fi module through Azure Logic Apps a serverless integration platform with over 1,400 prebuilt connectors into a Microsoft Fabric Event Stream. This high-throughput pipeline ingests data with sub-second latency, persisting it in Microsoft Fabric Eventhouse: a KQL-based columnar database purpose-built for time-series IoT workloads. Every bed state change is captured, timestamped, and immediately available for live querying in Fabric Real Time dashboards, power bi historical trend analysis, and predictive modelling.
Live Operational Intelligence: Dashboards That Drive Decisions
Data without visibility drives nothing. The framework surfaces live operational intelligence through Microsoft Fabric Real-Time Intelligence (RTI) into Power BI dashboards that refresh within seconds of events occurring on the ward floor giving managers, coordinators, and executives the situational awareness to act before problems escalate.
Live Bed Status
Real-time occupancy heat maps across every ward and bed type. Colour-coded status indicators for occupied, vacant, pending clean, and reserved.
LOS Trend Analysis
Length-of-Stay patterns by ward, specialty, and cohort with predictive discharge timing drawn from historical Eventhouse data.
Utilisation KPIs
Bed turnover rates, housekeeping SLA compliance, vacancy duration tracking, and occupancy benchmarks against operational targets.
The Real-Time Activator layer monitors the event stream and dispatches role-specific notifications: housekeeping supervisors receive an immediate Teams alert when a bed requires cleaning; charge nurses are notified when placements are confirmed; on-call coordinators are escalated to when capacity thresholds are breached. Automated Outlook summaries support shift handover briefings. The system closes the loop between intelligence and action across every operational role.
Measurable Business Impact
The Acuvate Healthcare Bed Management Framework delivers outcomes that are visible in the metrics healthcare leaders are held accountable for. The following value propositions reflect improvements observed across customer implementations:
15–20%
Reduction in Average Length of Stay
30%+
Faster Bed Turnaround Time
25%
Reduction in Manual Coordination Effort
HAI
Risk via Clinically Matched Placement
< 5s
Bed Reservation & Notification Cycle
24 / 7
Autonomous Agent Operation
Beyond operational metrics, customer organisations consistently report a secondary benefit that does not appear on a dashboard: staff relief. When coordinators are freed from reactive firefighting the constant phone calls, manual status updates, and cross-system checks they redirect their expertise toward complex, high-judgment patient care decisions. This translates into measurable improvements in staff satisfaction and retention, both of which carry significant financial consequences in a constrained clinical labour market.
Sensor Reliability and Change Management
Any IoT-dependent platform is only as dependable as its hardware layer. The Acuvate framework is engineered with a resilience-first approach to sensor infrastructure ensuring that hardware failures or signal inconsistencies never compromise patient placement decisions or operational continuity.
Redundant Sensing Logic
Each bed node uses both pressure sensor and RFID signals as independent confirmation channels. If one signal is absent or anomalous, the system flags the discrepancy and routes the decision to a human supervisor rather than proceeding on incomplete data.
Automated Fault Detection
The Fabric Event Stream monitors signals from every edge device. A missed signal triggers an immediate alert to the biomedical engineering team via Teams, with the affected bed automatically removed from the available pool until the fault is resolved.
Graceful Degradation
In the event of a sensor outage, the framework seamlessly falls back to EHR-sourced status data, maintaining operational continuity while engineering teams address the hardware issue. Coordinators are notified of degraded mode status in real time.
Change management for the sensor estate is governed through a structured lifecycle programme. All sensor nodes are registered in a centralised asset registry with firmware version tracking, maintenance schedules, and warranty records. Firmware updates are deployed over-the-air during low-activity windows with automatic rollback if a deployment introduces instability. Planned maintenance cycles are coordinated with ward management to ensure zero disruption to patient care activities. This operational discipline ensures that the IoT layer remains a reliable foundation as the framework scales across additional wards and sites.
Seamless EHR Integration and Interoperability
A bed management framework that operates in isolation from the clinical record is of limited value. The Acuvate Healthcare Bed Management Framework is designed from the ground up for deep, bidirectional integration with existing Electronic Health Record systems without requiring organisations to replace, significantly modify, or re-architect their core clinical platforms.
Standards-First Integration Architecture
The framework supports HL7 v2, HL7 FHIR R4, and REST API-based integration patterns, ensuring compatibility with all major EHR platforms including Epic, Cerner (Oracle Health), Meditech, and regional/national systems. Azure Logic Apps acts as the integration middleware, providing a low-code orchestration layer that maps, transforms, and routes clinical data between systems without bespoke point-to-point development.
The framework’s EHR integration covers the following key operational touchpoints:
- Admission and Transfer Events —Patient admission, ward transfer, and discharge events are consumed in real time from the EHR via HL7 ADT (Admit, Discharge, Transfer) message feeds, triggering the relevant agent workflows automatically without manual re-entry.
- ADT A01 (Admit), A02 (Transfer), A03 (Discharge) message types are natively supported.
- ADT A01 (Admit), A02 (Transfer), A03 (Discharge) message types are natively supported.
- Patient Demographics and Clinical Context — The Bed Recommendation Agent pulls structured clinical data from the EHR diagnosis codes, isolation flags, acuity scores, and care pathway information — to ensure bed matching decisions are grounded in the current clinical record.
- FHIR Patient and Encounter resources are used for structured data retrieval where supported.
- FHIR Patient and Encounter resources are used for structured data retrieval where supported.
- Bed Status Write-Back — Once a bed is reserved and a patient placed, the framework writes the updated bed assignment back to the HER along with Fabric RTI, maintaining a single source of truth for clinical staff and eliminating dual-entry.
- Configurable conflict resolution ensures EHR remains the authoritative record of care.
- Configurable conflict resolution ensures EHR remains the authoritative record of care.
Deployment flexibility is a core design principle. The framework supports cloud-native, hybrid, and on-premises EHR configurations, and can operate alongside existing bed management modules rather than replacing them allowing organisations to adopt the Agentic AI layer incrementally while protecting prior technology investments.
Data Governance and PHI Protection
In any healthcare AI deployment, data governance and the protection of Protected Health Information (PHI) are not optional features they are foundational design requirements. The Acuvate Healthcare Bed Management Framework is architected to comply with HIPAA, and to meet the data residency and privacy obligations applicable across international healthcare jurisdictions including NHS Digital Standards, GDPR, and regional healthcare data regulations.
PHI Minimisation
The framework operates on the minimum necessary PHI principle. Sensor telemetry carries only anonymised bed and device identifiers. Patient identity is resolved within secure, access-controlled service boundaries and is never transmitted in plaintext through the IoT pipeline.
Encryption at Rest & in Transit
All data in Microsoft Fabric Eventhouse is encrypted at rest using AES-256. All telemetry transmitted through Azure Logic Apps and the Event Stream pipeline uses TLS 1.3. PHI fields stored in Eventhouse are subject to column-level security policies.
Role-Based Access Control
Access to patient-identifiable data is governed through Microsoft Entra ID with role-based access control (RBAC). Clinical staff, housekeeping, and analytics users have access only to the data fields and dashboards relevant to their operational role.
Audit Logging
Every data access event, agent action, and system notification is written to an immutable audit log in Microsoft Fabric. Logs are retained in accordance with applicable regulatory retention schedules and are available for compliance reporting and incident investigation.
Data Residency
All data processing and storage occurs within the Microsoft Azure region specified by the customer, ensuring compliance with national data residency requirements. The framework supports geo-redundant configurations for disaster recovery without cross-border data transfer.
AI Transparency
Agent decisions are logged with the contextual reasoning that informed each recommendation. This transparency layer supports clinical governance requirements, enabling supervisors to review, audit, and where necessary override agent actions with a full decision trail.
Acuvate’s implementation engagement includes a dedicated Data Governance Assessment as part of the deployment programme. This assessment maps the organisation’s existing data classification policies, identifies PHI flows within the framework, and produces a Data Processing Agreement (DPA) and DPIA (Data Protection Impact Assessment) tailored to the customer’s regulatory context. Security controls are validated against the customer’s internal information security standards prior to production go-live.
Scaling Across the Enterprise
The framework is architected for enterprise-scale deployment from day one. Initial implementations typically begin with a defined scope a single ward or clinical unit to establish baseline metrics and validate configuration against the organisation’s operational environment. Once the value case is established, typically within eight to twelve weeks, the platform extends across additional dimensions without architectural rework:
- Additional agent configurations for theatre recovery beds, ICU step-down transitions, and elective surgery pre-admission scheduling.
- Expanded IoT coverage incorporating infusion pump telemetry, ventilator status, nurse call integration, and environmental sensors.
- Multi-site deployment across hospital networks with centralised governance and site-specific operational configuration.
- Extended EHR integrations as the organisation’s clinical system landscape evolves, supported by Logic Apps’ library of prebuilt connectors.
Every component of the framework — Microsoft Fabric, Azure Logic Apps, Copilot Studio, and the Agentic AI orchestration layer — is cloud-native and inherently elastic. Organisations already invested in Microsoft’s healthcare cloud ecosystem will find that the framework extends their existing infrastructure rather than displacing it, protecting prior investments while delivering transformative new capability.
The Next Evolution: Predicting Discharge, Not Just Managing Beds
The current framework excels at the operational layer ensuring the right bed is available for the right patient at the right time, with real-time visibility and automated coordination. But the most powerful lever for reducing length of stay lies upstream of the bed itself: in understanding, predicting, and actively managing each patient’s journey toward discharge from the moment they are admitted.
Acuvate’s next phase extends the framework with a Predictive Discharge Intelligence layer — an AI and machine learning capability that generates accurate Estimated Discharge Dates for every patient, identifies the barriers standing between each patient and a timely discharge, and equips care teams with the prioritised actions they need to reduce avoidable delays. Combined with workflow automation and a comprehensive analytics suite, this evolution transforms the framework from a bed management platform into a complete patient flow intelligence engine.
The Core Insight
Most excess days in hospital are not clinically inevitable — they are operationally avoidable. A patient waiting for a therapy assessment, a social care package, or a transport booking is not medically delayed; they are logistically delayed. AI-driven discharge planning surfaces these barriers early, when there is still time to act, rather than after the excess day has already been consumed.
Capability 1: Disposition Intelligence and Automation
Disposition Intelligence and Automation
AI/ML models that populate EDD and disposition automatically — from the first morning after admission
Machine learning models analyse each patient’s clinical profile, admission data, diagnosis codes, and care pathway on the first morning after admission automatically populating an Estimated Discharge Date and a predicted disposition (e.g., home, step-down, rehabilitation, long-term care). This eliminates the manual estimation burden from care teams and establishes a data-driven discharge target from day one. Critically, the models do not generate a single static prediction. They continuously update as the patient’s clinical status, test results, and care progress evolve — ensuring that the EDD and disposition intelligence remain accurate throughout the stay, not just at admission. The models also identify opportunities for earlier discharge and lower-acuity level-of-care transitions that clinical teams may not immediately recognise, creating initiative-taking capacity before it is urgently needed.
Auto-Populated EDD
EDD and disposition are automatically generated for every patient on the first morning post-admission requiring no manual entry from clinical staff.
Continuous Model Updates
Predictions refresh as clinical data evolves, keeping discharge targets accurate and actionable as the patient’s stay progresses.
Earlier Discharge Signals
Models flag opportunities for lower-acuity transitions and earlier discharges before they become obvious clinically creating capacity ahead of demand.
Capability 2: Care Progression Manager
Care Progression Manager
Structured care planning, MDR streamlining, and next-best-action guidance for every patient
The Care Progression Manager is the operational interface through which care teams interact with the framework’s discharge intelligence. It presents each patient’s EDD, disposition prediction, and identified barriers in a structured, actionable format enabling nurses, physicians, case managers, and social workers to evaluate the AI’s recommendations and drive discharge planning forward with a clear plan of action. The tool is specifically designed to streamline Multidisciplinary Discharge Rounds (MDRs). Rather than entering each MDR with an informal, memory-dependent discussion, care teams are equipped with a pre-structured patient-by-patient view showing predicted discharge dates, identified barriers, outstanding actions, and recommended next steps enabling rounds to be more focused, efficient, and clinically productive. For each patient, the system surfaces the single most impactful next best action: the one step that, if completed today, most advances the patient toward a timely discharge.
Barrier Identification
- Automatically surfaces pending clinical, social, and logistical barriers therapy assessments, imaging results, social care packages, transport bookings before they become day-of-discharge blockers.
- Each barrier is assigned a priority score based on its estimated impact on discharge timing.
MDR Streamlining
- Pre-populates the MDR agenda with structured patient-by-patient discharge intelligence, replacing informal whiteboard-based discussions with data-driven planning.
- Reduces average MDR duration and improves the consistency of discharge planning decisions across care team members.
Capability 3: Flow Prioritisation
Flow Prioritisation
ML models that determine the optimal sequence of clinical orders to maximise patient throughput
Even when discharge plans are well-formed, patient flow can be delayed by the sequencing of clinical orders and tasks. A single high-priority order that is not completed in time a physiotherapy assessment, a medication reconciliation, a radiology report can push a discharge from morning to afternoon, or from today to tomorrow. Flow Prioritisation uses machine learning models that leverage patient data, current census conditions, and order completion patterns to determine the optimal sequence in which care teams should complete outstanding orders to maximise patient flow. Rather than working through a list in arbitrary order, care teams are guided to the actions that will most directly remove barriers for patients who are nearing discharge — ensuring that high-impact tasks are completed on time, every time.
Demonstrated Clinical Outcome
Flow Prioritisation has delivered a 7% increase in on-time completion of high-priority clinical orders. This improvement directly correlates to 0.4 fewer excess days per patient a meaningful reduction in avoidable length of stay that, at scale across a hospital’s daily census, represents significant capacity recovery and cost avoidance.
Priority-Ordered Worklists
Every care team member sees a dynamically reordered task list that places discharge-critical actions at the top replacing intuition-based prioritisation with ML-driven sequencing.
Delay Prevention
By surfacing high-impact orders early enough in the clinical day to complete them, Flow Prioritisation prevents the late-afternoon discharge delays that consume beds into the following morning.
0.4 Fewer Excess Days
Validated reduction in excess days per patient attributable to improve on-time completion of high-priority orders creating measurable capacity at the whole-hospital level.
Capability 4: Insights Suite
Insights Suite
Comprehensive analytics for leadership accountability, outcome measurement, and continuous improvement
Operational intelligence is only transformative if it can be measured, interrogated, and acted upon at the leadership level. The Insights Suite is a comprehensive analytics platform designed specifically for hospital executives, operations directors, and quality improvement teams — providing the visibility needed to manage accountability, quantify return on investment, and identify the next highest-impact opportunity for improvement. Built on Microsoft Fabric and surfaced through Power BI, the Insights Suite provides real-time and longitudinal analytics across all dimensions of the framework: discharge prediction accuracy, LOS performance by ward and specialty, barrier type frequency, MDR adherence rates, flow prioritisation compliance, and bed turnaround trends. Leadership teams can drill from hospital-wide summaries to individual ward or patient-cohort performance — enabling both strategic oversight and targeted operational intervention.
Accountability Management
Ward-level and team-level discharge performance metrics with configurable benchmarks creating a structured accountability framework that leadership teams can review in regular operational forums.
Outcome & ROI Analysis
Tracks LOS reduction, excess day avoidance, bed utilisation improvement, and cost savings attributable to framework components providing the evidence base for continued investment and scale.
Opportunity Identification
Surfaces the specific wards, patient cohorts, barrier types, or workflow gaps where the largest untapped improvement opportunities remain guiding the next cycle of targeted intervention
The Time for Intelligent Bed Management Is Now
Hospital bed management has been a known, solvable problem for decades. What has changed is the availability of the technology required to solve it at the level of sophistication the challenge demands: AI agents that reason and act autonomously; IoT sensors that provide objective real-time ground truth; a streaming data platform that processes thousands of events per second; enterprise-grade security and governance controls that meet the stringent requirements of healthcare data regulation; and deep EHR interoperability that ensures the framework works with not around existing clinical systems.
Acuvate has assembled these capabilities into the Healthcare Bed Management Framework a production-deployed solution that is improving outcomes for patients, reducing operational costs for providers, and giving clinical staff back the time and cognitive space to focus on what matters most. The question for healthcare leaders is no longer whether intelligent bed management is achievable. It is whether your organisation will lead the transformation or follow it.
Deploy the Framework in Your Organisation
The Acuvate Healthcare Bed Management Framework is available for rapid deployment through a structured implementation engagement. Our healthcare technology and clinical operations teams work alongside your IT, digital transformation, and clinical leadership to configure, integrate, and validate the framework against your specific environment including full data governance assessment, EHR integration mapping, and staff change management support. Contact us at acuvate.com to schedule a solution demonstration and architecture review tailored to your organisation.
Healthcare Bed Management Framework - FAQs
An EHR-Integrated Bed Management System is a platform that syncs directly with clinical records (like Epic, Cerner, or Meditech) to manage patient flow. Without this integration, hospital staff must manually enter data twice, leading to “operational latency” where a bed is physically empty but the system still shows it as occupied. The Acuvate framework uses HL7 FHIR standards to ensure that when a patient is discharged in the EHR, the bed management system reacts in under 5 seconds, triggering cleaning and admitting the next patient immediately.
Traditional bed management in hospital settings is “passive,” meaning it shows a dashboard and waits for a human to act. Agentic AI in Healthcare is “active.” It uses autonomous agents to reason across clinical data (age, diagnosis, acuity) and IoT signals to:
- Recommend the optimal bed based on clinical fit, not just the first one available.
- Execute the reservation and notify housekeeping/nursing via Microsoft Teams without waiting for human initiation.
This shifts the staff’s role from manual coordination to high-level clinical oversight.
Real-time Hospital Bed Tracking provides the “ground truth” of what is happening on the ward. By utilizing Smart Hospital Infrastructure IoT—specifically Pressure Mat Sensors in mattresses and RFID Patient Wristbands—the system knows the exact second a bed becomes physically vacant. This data flows through Microsoft Fabric for Healthcare, eliminating the 2-to-4-hour delays common in manual systems and reducing the average length of stay (LOS) by 15–20%.
Yes. Predictive Hospital Capacity Management moves from reacting to empty beds to predicting when they will be empty. Using machine learning models, the framework identifies “operationally avoidable” delays—such as a patient waiting for a transport booking or a pharmacy script—days in advance. By surfacing these barriers early, the framework has demonstrated a clinical outcome of 0.4 fewer excess days per patient, effectively “recovering” hospital capacity without adding new beds.
The primary goal of Bed Management and Tracking in Hospitals is to reduce “reactive firefighting.” When coordinators are freed from constant phone calls and manual status updates, they can redirect their expertise toward complex patient care. Key measurable outcomes include:
- 30%+ Faster Bed Turnaround Time.
- 25% Reduction in Manual Coordination Effort.
- Lowered HAI (Hospital-Acquired Infection) Risk through clinically matched placements that ensure patients are in the most appropriate environment for their condition.
