Maintenance Alert Queueing With Multi-Criteria Fair Ranking
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current alert ranking systems for medical imaging devices primarily rely on priority, which fails to consider other critical factors such as deadline, service contracts, and the number of similar systems, leading to suboptimal prioritization and potential missed alerts, especially in scenarios where predictive models become outdated or hospitals have limited resources.
Innovation Solution
Implementing a credit-based fair queuing process that assigns discrete values to multiple criteria, including predictive model performance, deadline, and service contract details, to generate a ranked list of alerts, ensuring that alerts are prioritized based on a comprehensive evaluation of these factors, rather than solely on priority.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Ease of operation
If alerts are ranked solely by priority, then the ranking process is simple and fast, but critical factors such as deadline, service contracts, and system redundancy are not considered, leading to suboptimal alert prioritization
Solution Approach 1:
The alert ranking process is segmented into multiple independent criteria (priority, deadline, service contract level, system redundancy) that are evaluated separately and then combined. Each criterion is assigned a discrete value, and these values are multiplied to compute a comprehensive share for each alert, enabling detailed evaluation while maintaining computational efficiency.
Solution Approach 2:
The system transitions from a single-dimensional priority-based ranking to a multi-dimensional evaluation space by introducing additional criteria (deadline, service contract, redundancy). Each alert is represented as a point in this multi-dimensional space, and the credit-based fair queuing algorithm navigates this expanded space to determine optimal ranking.
2Reliability
If multiple criteria are considered in alert ranking, then alert prioritization becomes more accurate and comprehensive, but the complexity of the ranking system increases
Solution Approach 1:
The system changes the parameters of the ranking process by introducing discrete value assignments for each criterion and using multiplicative combination to compute shares. This parameter transformation enables comprehensive multi-criteria evaluation while maintaining computational tractability through the credit-based fair queuing mechanism.
Solution Approach 2:
The ranking system is designed to be self-regulating through the credit-based fair queuing algorithm, which automatically balances multiple criteria without requiring external intervention. The algorithm inherently manages the complexity of multi-criteria evaluation by using share computations and credit accumulation to produce balanced rankings.
3Reliability
If a comprehensive multi-criteria evaluation is performed, then more factors are considered for better decision making, but the computational resources and time required increase
Solution Approach 1:
The system applies partial evaluation in that not all alerts need to be fully evaluated with all criteria simultaneously. The credit-based fair queuing algorithm processes alerts in batches, accumulating credits over time, which allows for progressive evaluation rather than requiring complete evaluation of all alerts at once, reducing immediate computational burden.
4Quantity of substance
If alerts with low priority are included in the ranked list, then more alerts can be monitored, but high priority alerts may be overwhelmed and not reviewed promptly
Solution Approach 1:
The credit-based fair queuing algorithm creates equipotentiality in the alert evaluation process by ensuring that all alerts, regardless of priority level, have an opportunity to accumulate credits and reach the top of the ranked list. High priority alerts accumulate credits faster, but low priority alerts are not completely suppressed, creating a balanced evaluation environment that prevents any single alert from permanently dominating the queue.
Data Source
AI summary
A method (100) of generating a ranked list (42) of alerts (30) from files (32) of electronic devices (12) includes: applying predictive models (34) to data of the files to generate maintenance alerts for the electronic devices; assigning discrete values for a set of criteria (36) to each maintenance alert whereby the maintenance alert has a share computed as a product of the assigned discrete values; performing a credit-based queuing process (38) to assign the maintenance alerts to slots of a ranked list, wherein in the credit-based queuing process the maintenance alerts accumulate credits in accord with the shares computed for the respective maintenance alerts; and displaying, on a display device (24) operatively connected with the at least one electronic processor, the maintenance alerts as the ranked list in accordance with the assignments of the maintenance alerts to the slots of the ranked list.