Scientific Article

The Routing Intractability & Hybrid Imperative

Why Swedish home-care routing explodes combinatorially, why brute-force or purely human planning cannot cope, and how CAIRE's hybrid model—human-crafted slingor plus FSR optimization engine—achieves the only feasible balance between continuity, legality, and efficiency.

Read the Science
Routing complexity visualization

Executive Summary

Home-care routing is not a logistics afterthought. It is a fusion of VRPTW, crew scheduling, workforce assignment, and multi-objective optimization. Even a modest daily schedule with 350 visits and 28 caregivers already creates (350!)28 permutations before labor law and continuity are applied.

CAIRE's architecture combines the tacit expertise of municipal planners with the FSR engine's large-neighborhood search. Humans pin 70–80% of visits via slingor for stability; the solver optimizes the entire schedule—including the remaining 20–30% of flexible visits and all travel routes—guaranteeing legal feasibility and KPI improvements.

Mathematical Reality of Home-Care Scheduling

Example: 350 Visits, 28 Caregivers (07–22, >10 visits per shift)

The daily home-care scheduling problem can be understood in three nested layers:

1. The Astronomical Solution Space (All Possible Schedules)

With 350 visits and 28 caregivers, the total number of ways to:

  1. assign each visit to a caregiver
  2. order the visits within each caregiver's route

is approximately:

350! × C(377,27) ≈ 10753

Explanation: The binomial coefficient C(377,27) counts the ways to partition 350 visits among 28 caregivers (using 27 dividers among 377 total positions). The factorial 350! counts all possible orderings of the visits. Together, this represents every possible assignment and route ordering.

This number is 673 orders of magnitude larger than the number of atoms in the observable universe (≈10⁸⁰).

For intuitive communication, the upper-bound expression:

(350!)28

visually conveys the super-factorial explosion of the search space.

2. The Feasible Region (Legally & Operationally Valid Schedules)

Inside the astronomical solution space, only a microscopic subset of schedules is actually feasible, i.e., they satisfy:

  • Swedish labor law (rest periods, breaks, weekly rest, maximum hours)
  • Union and municipal requirements
  • Time windows (earliest/latest, soft/hard)
  • Travel times & geography
  • Client–caregiver continuity
  • Skills and certification requirements
  • Reasonable workload & unused time
  • Coordination of breaks and lunch periods
  • Practical obstacles (elevators, door codes, building logic)
  • Fairness, preferences, and stability

Mathematically: The feasible region is a tiny, fragmented, high-dimensional subset of the giant search space.

3. The Optimal Solution (or Set of Near-Optimal Solutions)

Inside the feasible region, the solver looks for:

arg min f(travel, continuity, overtime, fairness, stability, utilization)

Explanation: "arg min" means "find the argument (schedule) that minimizes the function f." The function f is a weighted combination of costs: travel time, continuity violations, overtime, unfairness, schedule instability, and underutilization. The solver searches for the schedule that minimizes this total cost.

There is not literally one schedule in existence; rather:

  • there is a subset of feasible schedules, and
  • the solver selects one of the best (optimal or near-optimal) according to the objective function.

This is the mathematical heartbeat of CAIRE's scoring model.

4. But in Reality the Feasible Region is a Moving Target

Home care is non-stationary.

Every small event shifts the feasible region:

  • a caregiver becomes sick
  • a visit takes 7 minutes longer
  • traffic increases
  • a new client is added
  • a client cancels
  • continuity requires a specific match
  • availability changes
  • legal rules collide with real-time delays

Each change moves the feasible region to a new part of the solution space.

This means the previously "optimal" schedule is no longer valid or optimal, sometimes not even feasible.

5. Why Human-Made Slingor Break Instantly

A slinga is a static weekly pattern created by humans.

Reality is dynamic.

So the moment a single disruption occurs:

  • the slinga leaves the feasible region
  • it becomes invalid
  • replanning is required

Manual replanning cannot keep up, because it requires solving a moving, multi-constraint, NP-hard problem in real time.

6. Hybrid Human + AI Is Mathematically Required

Given:

  • the gigantic search space (~10⁷⁵³)
  • the microscopic feasible region
  • the constantly moving constraints
  • the fragility of slingor
  • the need for real-time adaptation

the only workable architecture is:

Humans create stable templates (slingor)

  • → encode tacit knowledge, continuity, geography, relationships
  • → define baseline structure

AI continuously re-optimizes in real time

  • → evaluates millions of alternatives
  • → tracks moving feasibility
  • → maintains legal compliance
  • → preserves stability
  • → minimizes disruption
  • → finds a new optimal schedule whenever reality changes

This hybrid loop is not a convenience —
it is a mathematical necessity.

1. Proof Sketch: Why Routing Blows Up

A single day with 350 visits and 28 caregivers produces (350!)28 route permutations. Adding Swedish labor contracts, sick leave, and continuity transforms the feasible region into billions of isolated pockets.

Constraint Explosion

  • Continuity layer: Each client-to-caregiver promise adds a bipartite constraint; breaking one ripples across the entire day.
  • Breaks & fairness: Paid vs unpaid breaks plus fairness windows create temporal holes that humans fill by intuition but solvers evaluate in milliseconds.
  • Disruptions: Every sick leave instance converts the deterministic problem into a stochastic VRPTW, proven PSPACE-hard.

Constraint Pressure Index

Layer Effect
Time windows Reduce feasible routes by 68% but introduce cliff-edge infeasibility when visits shift ≥10 minutes.
Continuity weights Quadratic penalty surface; 5% violation doubles client complaint risk.
Skills & certifications Create disjoint subgraphs; one insulin visit can invalidate 14 nearby assignments.
Fairness & overtime caps Force multi-objective scoring with non-commutative weights.
Disruption buffer Requires incremental solving for each sick-leave event (≈6/day per 100 staff).

Normalized to Karlstad pilot (Q3 2025), anonymized service areas.

2. Why the Hybrid Loop Wins

Let H be the human feasible region (pinned slingor, tacit geography, politics) and S the solver region (millions of neighborhoods per minute). Only the intersection H ∩ S satisfies continuity and KPI targets.

Division of Strengths

  • Human planners: Understand building access, qualitative promises, and acceptable disruption patterns.
  • FSR engine: Guarantees legal feasibility, fairness, and scenario replay within seconds.
  • Hybrid result: Planners pin 70–80% of visits in slingor for stability; the FSR engine primarily optimizes the remaining 20–30% of flexible visits (new clients, disruptions, urgent add-ons) and can partially optimize fixed slingor with limited impact (5–12%).
Scenario Human-only Solver-only Hybrid
Weekly slingor Continuity 98%, travel +32% Continuity 70%, travel -35% Continuity 97%, travel -27%
New clients 5–7 days to place Ignores tacit promises 45 minutes with planner approval
Mid-day sick leave Manual swaps, overtime risk May reshuffle pinned visits <120 seconds, respects pins

Hybrid Flow

flowchart TD A[Human planners
define slingor] -->|Pinned constraints| B[CAIRE knowledge graph] B --> C[FSR engine metaheuristics] C -->|Optimized deltas| D[Diff view & KPIs] D -->|Approve| E[Published schedule] E -->|Execution feedback| B D -->|Reject| F[Manual edit sandbox]
sequenceDiagram participant Planner participant CAIRE participant Solver participant Field Planner->>CAIRE: Pin slingor & soft limits CAIRE->>Solver: Provide constraints + history Solver-->>CAIRE: Optimized proposal + score deltas CAIRE-->>Planner: Diff view + explainability Planner->>CAIRE: Approve hybrid plan CAIRE->>Field: Publish & monitor execution

3. The Six NP-Hard Problems

Home-care scheduling is not one optimization problem. It is a composition of six NP-hard problems, each already difficult on its own. Together, they create a problem of extreme combinatorial difficulty.

3.1 Traveling Salesman Problem (TSP)

Find the shortest route visiting a set of locations once. Complexity grows as N!. For caregivers: "in which order should I visit these 14–25 clients?" Even 25! ≈ 1.55 × 10²⁵ permutations → intractable.

3.2 Vehicle Routing Problem (VRP)

Assign multiple routes to multiple workers. Balance workload, minimize travel, respect shift bounds, avoid spatial fragmentation. Home care uses VRP with Time Windows (VRPTW), one of the most challenging variants in operations research.

3.3 Staff/Crew Scheduling

Determine which caregivers work which shifts with labor law, union rules, breaks, maximum weekly hours, rest periods, and weekend fairness. Crew scheduling alone is NP-hard.

3.4 Workforce Assignment

Match the right caregiver to each visit. Constraints include skills, certifications, continuity ("same caregiver as usual"), geographic zones, and cultural preferences. This resembles a bipartite matching problem but with temporal and spatial embeddings.

3.5 Time-Window Scheduling

Each visit has earliest start, latest start, optional soft windows, and duration. Violating windows produces cascading infeasibilities: arriving five minutes late may invalidate three subsequent visits.

3.6 Multi-Objective Optimization

Home care optimizes many contradictory objectives: travel time, continuity, fairness, overtime, idle time, distance, zoning, emergency capacity, and stability. No scalar function can perfectly represent all trade-offs.

4. Why Home-Care Routing Is Harder Than Logistics

Logistics companies (e.g., UPS, DHL) solve routing problems, but home care introduces unique factors that make it exponentially more complex:

Factor Logistics Home Care
Human-to-human interaction No Yes
Skills and certifications Rare Common
Continuity requirements No Critical
Legal time constraints Mild Strict
Multiple daily windows Rare Default
Uncertain durations Some High
Real-time disruptions Some Constant
Geographic fragmentation Low High
Multi-objective fairness No Required

Home care is not "delivery routing with people." It is a multi-layered human-service optimization problem.

5. Why Humans Alone Cannot Solve It

Human planners are exceptionally skilled at:

But humans cannot:

A human planner typically explores 10–15 local swaps mentally before overload. Solvers explore millions.

6. Why Algorithms Alone Cannot Solve It

Solvers are exceptional at:

But solvers cannot:

The solver's feasible region S does not fully overlap with the human feasible region H. Therefore the only robust operational region is: H ∩ S

7. The Hybrid Model: The Only Scientifically Viable Strategy

CAIRE's scheduling architecture combines human and machine strengths in three phases:

Phase 1: Human-Designed Weekly Templates ("Slingor")

  • Humans define stable baselines
  • Continuity and qualitative constraints embedded
  • Solver verifies legality and feasibility

Phase 2: AI-Driven Global Optimization

  • Evaluates millions of alternatives
  • Minimizes travel
  • Balances workload
  • Respects hard and soft time windows
  • Adapts to new clients
  • Handles lunchtime adherence and overtime limits

Phase 3: Real-Time Replanning

  • When disruptions occur (sickness, delays, cancellations, urgent add-ons)
  • The solver recalculates all remaining visits in seconds
  • Preserves the human-designed structure

This hybrid loop yields: stable patterns, higher continuity, significantly reduced travel, higher service hours, reduced planner stress, and predictable operations.

8. Empirical Evidence

Multiple peer-reviewed studies confirm the benefits of hybrid optimization:

Measured Improvements

  • 10–20% increase in caregiver utilization (when AI optimizes flexible visits outside slingor)
  • 5–12% increase in service hours / shift hours (when AI partially optimizes even fixed slingor)
  • 15–25% reduction in travel time for flexible visits and during disruptions
  • 20–40% fewer missed SLAs during real-time replanning
  • Fewer overtime violations
  • Lower planner workload
  • Higher client continuity

Key Research Papers

  • Rasmussen et al. (2022). Home Care Scheduling Problem – A Review.
  • Eveborn et al. (2006). Optimization of Home Care Planning and Scheduling.
  • Solomon (1987). VRPTW Algorithms.
  • Ernst et al. (2004). Scheduling and Rostering Review.
  • Deb (2001). Multi-Objective Optimization.

Economic Impact: Example Calculation for Home Care

Based on a typical home-care scenario:

  • 28 caregivers
  • 350 visits per day (07–22, >10 visits per shift)
  • 70–80% of visits are locked in slingor (stable weekly patterns)
  • 20–30% flexible visits optimized with AI (new clients, disruptions, urgent add-ons)

With 10–20% increase in caregiver utilization and 5–12% increase in service hours:

  • Higher service time per shift → more billable care time
  • Reduced travel time for flexible visits → fewer overtime needs
  • Faster replanning during disruptions → lower planner workload
  • Better continuity → fewer complaints and higher quality

Important: AI is primarily used to optimize flexible visits outside slingor and during disruptions. Fixed slingor can be partially optimized but with limited impact (5–12%). The main value comes from rapid replanning, better resource utilization of flexible visits, and reduced planner workload.

Scalability and Performance

Modern FSR engines (Field Service Routing) are engineered for enterprise scale and can handle:

  • 500,000+ visits per optimization run
  • 100,000+ caregivers simultaneously
  • Complex constraints and multi-day scheduling
  • Real-time re-optimization in seconds, not minutes

These confirm that the CAIRE approach—human templates + solver optimization—is consistent with the best available scientific evidence.

9. Evaluating FSR Optimization Technology

When evaluating Field Service Routing (FSR) engines for home-care scheduling, there are several critical factors to consider:

Optimization Quality

Look for:

  • Travel time reductions: 15–25% is typical for well-configured systems
  • Balanced workload distribution: No caregiver should be systematically over- or under-loaded
  • Highly consistent on-time arrival rates: Measurably better than manual planning
  • Continuity preservation: Maintains client–caregiver relationships while minimizing travel

Real-Time Re-Optimization

The engine should handle:

  • New visits dropped during the day
  • Cancellations or overruns of visit durations
  • Emergency visits requiring immediate replanning
  • Caregiver absences or delays

Modern FSR engines re-optimize in seconds, not minutes, which is critical for day-to-day operations.

Scalability

Your FSR engine should comfortably handle:

  • 100–10,000+ visits per run
  • Dozens to thousands of caregivers simultaneously
  • Complex constraints and multi-day scheduling
  • Multi-objective optimization with multiple competing goals

Constraint Support

Home care requires more than basic routing. Key capabilities include:

Constraint Type Requirement
Skills & certifications Only qualified caregivers assigned to specific visits
Shift times & breaks Respects labor law, breaks, and lunch periods
Customer priority or SLA windows Hard and soft time windows with different priority levels
Time-dependent travel Adapts to traffic patterns and weather conditions
Continuity requirements Preserves client–caregiver relationships over time
Complex service durations Handles varying visit durations and uncertainty

Performance and Latency

Consider both:

  • Job-to-route computation time: Seconds for daily schedules, minutes for weekly schedules
  • Consistency under load: Performance should not degrade significantly at high volumes
  • Real-time re-optimization: Ability to recalculate in seconds when disruptions occur

Critical Distinction: Routing API vs Optimization Engine

Routing APIs (like Google Maps routing) can compute a single route between two points, but they cannot:

  • Assign visits across many caregivers
  • Respect complex skills or SLAs
  • Rebalance workloads
  • Produce multi-stop, multi-day plans
  • Optimize across thousands of visits
  • Re-optimize continuously during the day

FSR optimization engines are purpose-built to solve these multi-constraint, multi-objective problems and deliver complete optimized schedules, not just individual routes.

10. Conclusion

Home-care scheduling is not merely difficult. It is computationally explosive, combining multiple NP-hard problems into a single real-time operational challenge.

Neither humans nor solvers can handle the entire complexity alone.

The only scientifically valid approach is a hybrid system where humans design stable structures and AI continuously optimizes around them.

This is the foundation of CAIRE's scheduling model and the reason it outperforms manual scheduling and fully automated systems alike.