Enterprise AI Analysis
Conventional and hybrid time series models for forecasting medication dispensing and errors integration in automated dispensing cabinets
This study applies conventional, hybrid time series, and machine learning models to forecast key performance indicators of Automated Dispensing Cabinets (ADCs): items dispensation, override occurrences, and error integration. Using monthly data from a MICU at Almoosa Hospital, the research demonstrates that the NPAR-ANN hybrid model exhibits superior performance, achieving the lowest RMSE values for forecasting these critical ADC parameters. The findings provide data-driven insights to optimize medication management and decision-making in hospital settings.
Executive Impact: Key Performance Indicators
The NPAR-ANN hybrid model demonstrates significant improvements in forecasting critical ADC metrics, enabling proactive decision-making and enhanced patient safety. Here's a snapshot of its superior performance:
0
Lowest RMSE for Items Issued
0
Lowest RMSE for Overrides
0
Lowest RMSE for Error Integration
0
Accuracy Improvement Over Traditional Models
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Automated dispensing cabinets (ADCs) are a critical innovation in modern healthcare, improving medication management efficiency, accuracy, and security. They control medication storage, dispensing, and delivery, often integrating human operations with electronic information to track medication-related issues throughout the process. ADCs enhance patient well-being, improve workflow efficiency, and reduce healthcare costs.
97%
US Hospitals using ADCs
Despite significant benefits, including documented reductions in medication errors in ICUs and emergency departments, ADCs face challenges such as staff resistance and noncompliance with new workflows. Most previous studies focused on ADC adoption and safety improvements, neglecting predictive modeling approaches for dispensing volumes and integration errors.
This study utilized retrospective monthly time series data from January 2023 to December 2024 from the MICU at Almoosa Specialist Hospital. The dataset included three key performance indicators: items issued, override occurrences, and integration errors.
Hybrid Model Construction Process
Data Preprocessing
→
Select ARIMA Parameters
→
Fit The ARIMA Model
→
Generate ARIMA forecasts and calculates residuals
→
Train ANN on ARIMA Residuals
→
Train the network to learn the nonlinear structure in the ARIMA residuals
→
Combine ARIMA and ANN Outputs
→
Model Evaluation
| Statistics | No of items | Overrides | Integration errors |
|---|---|---|---|
| Minimum | 3621 | 60 | 9 |
| Maximum | 10686 | 194 | 182 |
| Mean | 7749.79 | 115.16 | 59.12 |
| SD | 2109.58 | 36.85 | 49.86 |
| ADF statistic | -3.19 (0.11) | -2.10(0.53) | -1.86 (0.62) |
The ADF test results indicated that all three series were non-stationary and required differencing (second-order for items and errors, third-order for overrides). Johansen cointegration tests showed no significant long-term relationships, justifying the differencing approach. Seasonal ANOVA found no meaningful seasonal patterns.
The NPAR-ANN hybrid model consistently outperformed all other tested models across all three performance indicators: items issued, overrides, and integration errors. It achieved the lowest RMSE values, demonstrating superior accuracy and robustness in capturing both linear and nonlinear patterns.
| Model | RMSE (Items Issued) | RMSE (Overrides) | RMSE (Integration Errors) |
|---|---|---|---|
| NPAR-ANN | 71.50 | 15.43 | 20.92 |
| LOESS (NPAR) | 502.75 | 36.68 | 31.89 |
| ESM | 2063.58 | 25.27 | 61.57 |
| ARIMA | 2247.52 | 38.32 | 60.44 |
| ANN | 3154.97 | 39.87 | 33.44 |
|
|||
Short-term forecasts (2 years) predict fluctuations in items dispensed, with peaks in mid-2025 and dips in mid-2026. Overrides are expected to rise gradually in 2025 before decreasing in 2026, suggesting improved system performance. Integration errors are forecasted to increase early 2025, then stabilize through 2026, indicating the potential for further system optimization.
This novel study modeled key parameters of ADCs, providing data-driven insights that can inform hospital decision-making and optimize medication management. The study showcased the application of hybrid machine learning models in forecasting critical ADC parameters, offering valuable data-driven insights for hospital administrators.
The main limitation is the focus on a single regional hospital, potentially limiting the generalizability of findings across different operational environments. The small sample size also means some models like ANN may not perform optimally without larger datasets. Future work should gather additional data from other regions to validate the model and explore more advanced forecasting models like Prophet or Long Short-Term Memory (LSTM) networks for complex healthcare time series.
Addressing Generalizability
Problem: Limited scope to a single regional hospital might not reflect diverse ADC performance across different healthcare settings.
Solution: Recommendation to gather additional data from other regions for validation, allowing for broader application and improved decision-making policies.
Outcome: Improved model robustness and applicability, guiding hospitals to enhance their medication management practices more effectively.
Calculate Your Potential ROI
Estimate the financial and operational benefits of implementing advanced AI solutions in your enterprise.
Estimated Annual Savings
$0
Annual Hours Reclaimed
0
Your AI Implementation Roadmap
We guide you through a structured process to seamlessly integrate AI into your operations and unlock its full potential.
Phase 1: Discovery & Strategy
Comprehensive assessment of your current workflows, identification of AI opportunities, and development of a tailored strategy aligned with your business objectives.
Phase 2: Solution Design & Development
Custom AI model design, data preparation, and iterative development of the solution, ensuring it meets your specific requirements and integrates with existing systems.
Phase 3: Deployment & Integration
Seamless deployment of the AI solution into your enterprise environment, including API integrations, infrastructure setup, and rigorous testing for performance and security.
Phase 4: Optimization & Scaling
Continuous monitoring, performance tuning, and scaling of the AI solution to maximize its impact and adapt to evolving business needs, ensuring long-term value creation.
Ready to Transform Your Enterprise with AI?
Book a free consultation with our AI specialists to explore how our solutions can drive efficiency, accuracy, and innovation in your organization.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat