Clever radiology workflow optimization with AI brokers

Many healthcare organizations report that conventional worklist programs depend on inflexible guidelines that ignore vital context, radiologist specialization, present workload, fatigue ranges, and case complexity. This creates a persistent problem: radiologists cherry-pick simpler, higher-value instances whereas avoiding advanced research, resulting in diagnostic delays and elevated prices. Analysis throughout 62 hospitals analyzing 2.2 million research discovered that inefficient case project causes 17.7-minute delays for expedited cases and costs of $2.1M–$4.2M across hospital networks. The basis trigger is easy: conventional radiology worklist programs depend on inflexible, rule-based engines that ignore the context that issues most — radiologist specialization, present workload, fatigue ranges, and case complexity. On this submit, we’ll present methods to construct an radiology workflow optimization with AI brokers on Amazon Bedrock AgentCore and Strands Agents SDK .

Radiologist worklist programs depend on deterministic, rule-based engines that route research in accordance with predefined logic. Static specialty matching ignores context, reminiscent of whether or not the out there radiologist has been decoding advanced instances for a number of consecutive hours or whether or not an easy follow-up scan really warrants subspecialist consideration. Workload balancing responds to present queue depth slightly than anticipating calls for primarily based on case complexity, estimated interpretation time, or doctor fatigue patterns. Most critically, no studying happens when deterministic guidelines produce suboptimal assignments, the identical inefficient patterns repeat till somebody manually updates the underlying logic. On this submit, you may discover ways to:

Scale back diagnostic delays by constructing an clever worklist system
Deploy AI brokers that motive about your crew’s specialization, workload, and fatigue
Implement context-aware case project that reduces diagnostic delays

By transferring past inflexible, deterministic guidelines towards Agentic AI that really understands our subspecialties, we’re witnessing a paradigm shift that elevates radiology workflow from easy activity administration to really autonomous orchestration. The correct subspecialist is seamlessly matched with the precise case on the proper time, liberating radiologists to focus fully on diagnostic excellence slightly than navigating the queue. Radiology Companions acknowledges this as a mission-critical workflow functionality and is partnering with AWS to undertake Agentic AI for clever workflow optimization.

Agentic AI strategy

An AI agent is an autonomous software program element that may understand its setting, motive about objectives, and take actions to attain them. In your radiology workflow optimization, a community of specialised AI brokers collaborates to orchestrate advanced medical workflows from begin to end. Every agent handles particular duties inside the workflow. Brokers coordinate throughout specialties and adapt to ship optimum outcomes for sufferers and crew. AI brokers on Bedrock AgentCore consider a number of elements concurrently reminiscent of radiologist specialization, present workload, fatigue patterns, case complexity, medical urgency, and availability to make optimum case assignments. The AI fashions powering the brokers are basis fashions (FMs) out there via Amazon Bedrock. The system constantly learns from historic patterns and adapts to altering situations, minimizing the motivation constructions that drive cherry-picking conduct.

Overview of the answer

This part walks you thru the answer structure and implementation of accelerating radiology imaging workflows by intelligently optimizing examination prioritization and radiologist project. A pattern examination project output from the clever worklist orchestrator is proven within the following determine. A knee MRI examine arrives in image archiving and communication system (PACS) and must be assigned. The agentic worklist optimization system suggests the first project together with rationale as beneath.

The answer structure exhibits elements described within the following sections.

1. The workflow is initiated when a technologist acquires a brand new examination that turns into out there within the image archiving and communication system (PACS) for studying. A queue of exams verified by technologists for picture high quality await project to the very best out there radiologist. The project course of operates as an asynchronous workflow, the place exam-to-radiologist matching triggers primarily based on dynamic guidelines. The objective of the system is to assign the precise radiologist to the precise examination on the proper time.
2. The examination project set off initiates AgentCore Runtime session by calling Clever worklist orchestration agent (2), which represents the mind of the answer. The orchestration agent is answerable for coordinating a number of specialised AI brokers that execute their respective duties in parallel. For routine workflows, the orchestrator first coordinates with two brokers, the Examination Metadata Synthesizer and Affected person Historical past Synthesizer to gather related contextual data. Based mostly on this aggregated knowledge, the Rad Task Agent applies reasoning logic to match the examination with the optimum radiologist. For precedence instances, triaging programs establish vital findings requiring speedy consideration. When AI algorithms detect pressing situations reminiscent of intracranial hemorrhage, they robotically set off examination prioritization, prompting the orchestrator to flag a high-priority indicator for the studying radiologist. The brokers are hosted on AgentCore Runtime, utilizing the AgentCore Runtime starter toolkit, the AgentCore SDK or immediately via AWS SDKs.
3. Amazon Bedrock Guardrails is utilized at two factors within the worklist stream. On the inbound facet, it intercepts queries earlier than they attain the Worklist orchestrator, rejecting prompts that try to extract affected person personally identifiable data (PII), reminiscent of names, SSNs, addresses from the medical knowledge shops. On the outbound facet, it scans agent responses from the Examination metadata, Scientific knowledge historical past, Rad mapper, Examination prioritization and Dynamic guidelines brokers to redact PII that will have surfaced throughout retrieval from AgentCore Reminiscence or the Scientific knowledge API. This fashion, brokers internally function on full exam-level knowledge for correct optimization, however solely floor operationally related data (examination sort, modality, urgency, scheduling) again to the consumer. Matter restrictions additional constrain brokers to worklist optimization queries solely.
4. The Examination metadata synthesizer agent (3a) extracts examination particulars together with modality, physique half, and urgency flags from incoming research. Concurrently, the Affected person historical past synthesizer agent (3b) gathers related medical context and retrieves prior examination information to supply complete affected person background data that informs prioritization selections.
5. The Rad Task Agent (4) optimizes radiologist allocation for every examination by analyzing a number of elements together with radiologist profiles, roles, specialties, most popular hospital affiliations, real-time availability, and dynamic enterprise guidelines. The agent intelligently balances the worklist by matching every examine to the radiologist whose specialization aligns with the examination sort, prioritizing STAT instances to fulfill pressing necessities, and distributing a wholesome mixture of advanced and routine research to stop fatigue. Future enhancements can allow the agent to route research primarily based on their originating hospital and corresponding Service stage settlement (SLA) turnaround time necessities.
6. The Rad availability sub agent (4a) checks real-time schedules and present workload distribution to stability case allocation. Moreover, the Dynamic guidelines agent (4b) applies important enterprise logic together with service stage settlement necessities, new modalities and examination varieties, and escalation insurance policies for compliance with institutional and contractual obligations. The agent may even use unstructured notes from the technologist in determination making for matching.
7. AgentCore Reminiscence maintains contextual data for examination processing via two complementary reminiscence programs:
  - Brief-term Reminiscence shops uncooked interactions to protect context inside particular person classes. It captures the entire dialog historical past as sequential occasions, with every examination metadata entry and agent response saved individually. This structure helps the agent to reconstruct the complete dialog historical past, sustaining continuity even after service restarts or examination reprioritization triggers. When an assigned examination fails to fulfill its service stage settlement (SLA), a set off notifies the orchestrator to provoke the reassignment. The system retrieves examination metadata from short-term reminiscence context and invokes solely the radiologist availability agent. Equally, when an assigned radiologist rejects or skips an examination, the reassignment course of is robotically triggered primarily based on short-term reminiscence context for accelerated project.
  - Lengthy-term reminiscence gives persistent data retention throughout a number of classes utilizing a semantic reminiscence technique. The system extracts and shops key details about examination assignments, together with Order MRN (Medical Report Quantity) and assigned radiologist, process sort and imaging modality, affected person medical historical past, project rationale, and determination elements. This persistent data base maintains a complete radiologist project historical past, which helps the system be taught from previous selections and optimize future examination distributions primarily based on historic patterns, radiologist experience, and workload balancing. Whereas semantic reminiscence retains details, AgentCore’s episodic reminiscence captures experience-level data: the objectives tried, reasoning steps, actions taken (together with instruments used and context or parameters handed), outcomes, and reflections of the outcomes. As a substitute of storing each uncooked occasion, it identifies vital moments like SLA breaches or project rejections by radiologists, summarizes them into compact information, and organizes them so the system will retrieve what issues with out noise. Reflections rework episodic experiences into strategic data by figuring out patterns, extracting insights, and synthesizing actionable steering that helps brokers to be taught and make more and more knowledgeable selections over time.
8. Examination prioritization agent (5) will triage the exams utilizing imaging fashions that establish the necessity to improve the precedence of an examination primarily based on a vital discovering like acute pulmonary embolism, a situation that wants speedy consideration to optimize medical outcomes. This asynchronous workflow processes photographs via AI imaging fashions reminiscent of Artery-aware network (AANet) for pulmonary embolism detection in CT pulmonary angiography (CTPA) photographs. When fashions detect vital findings with excessive confidence, they robotically set off examine prioritization for speedy radiologist assessment.
9. As soon as the examination is assigned to a radiologist, they will work together with an clever front-end workflow administration utility that makes the workflow optimization accessible via a user-friendly interface. The radiologist can settle for, reject, or skip the project and proceed with studying. The radiologist’s selections are robotically discovered by the system to enhance over time. For instance, steady adaptive studying by analyzing suggestions loops and contextual judgment, the agentic system refines case distribution in real-time, decreasing the cognitive load on radiologists. Episodic reminiscence technique reflections constructed on episodic information like SLA breach, project rejection assist analyze previous episodes to floor insights, patterns, and higher-level conclusions. As a substitute of merely retrieving what occurred, reflections assist the system perceive why sure occasions matter and the way they need to affect future conduct.
10. When brokers require exterior knowledge to finish their duties, they invoke instruments by way of the /mcp endpoint via the AgentCore Gateway. This gateway serves because the central integration hub for the complete structure, dealing with Mannequin Context Protocol (MCP) routing together with inbound and outbound authentication for system communications. The gateway connects to AgentCore Identification, which integrates with exterior id suppliers for safe entry management throughout system interactions and knowledge exchanges.

Instrument requests are routed to the MCP Server inside the AgentCore Runtime, which exposes a number of backend instruments important to the workflow. These built-in instruments embody entry to Scientific knowledge API for accessing affected person information and medical histories from digital well being file (EHR) programs and the Rad calendar for retrieving radiologist scheduling data via MCP server. The instruments will use present enterprise Imaging APIs for direct imaging examine entry from PACS by way of OpenAPI specs.

Implementation steps

The next steps are wanted to implement the answer. For the total code, see the GitHub repository.

The clever worklist orchestrator agent makes use of the agent-as-tool sample and has entry to 4 Strands instruments as sub-agent. The orchestrator agent determines which specialised “tool-agent” is finest fitted to a sub-task. It then “calls” that agent as if it had been a perform. When known as, the sub-agent takes over the sub-task. It makes use of its personal massive language mannequin (LLM) and immediate to motive via the issue, calling its personal instruments a number of occasions earlier than returning a synthesized outcome to the orchestrator. The agent makes use of its built-in MCP shopper to provoke communication to the precise instruments via the AgentCore Gateway. This permits the agent to execute advanced duties autonomously through the use of these instruments for real-world motion for matching radiologists primarily based on their specialties, retrieving affected person medical historical past, extracting examination metadata, and checking their shifts. This agent makes use of the next system immediate:

MAIN_SYSTEM_PROMPT = ""

You're a Radiologist Task Orchestrator Agent answerable for figuring out and recommending essentially the most acceptable radiologist for a brand new medical imaging examine.

You obtain a consumer question together with a JSON object containing related examine and affected person knowledge. 

Function & Tasks
Your main duties are:

Delegate particular duties to specialised sub-agents: rad_mapper, image_assessor, clinical_data_collector, metadata_finder, and shift_checker. 
Gather related historic affected person knowledge and collect detailed details about the imaging examine, significantly rom its metadata
Analyze all collected data to establish and return a prioritized checklist of acceptable radiologists for project
Handle the end-to-end workflow throughout all system elements
Be certain that all suggestions align with established medical finest practices

Instrument choice
All the time choose essentially the most acceptable sub-agent or device primarily based on the character of the incoming question and the info out there.

Behavioral Pointers
It's essential to all the time:

Keep HIPPA compliance and shield affected person knowledge privateness at each step
Observe established medical workflows with out deviation
Doc determination rationale clearly and transparently for each suggestion
Coordinate successfully with all sub-agents for seamless data stream
Prioritize affected person security above all different concerns in each suggestion

Output Format
Return the advisable radiologists in precedence order, together with a short rationale for every suggestion primarily based on the examine sort, metadata, affected person historical past, and radiologist availability/experience.
""

The MCP server makes use of FastMCP with stateless HTTP transport, exposing instruments adorned with @mcp.device() that present radiologist search, imaging examine metadata, affected person medical knowledge, and shift availability. These MCP instruments are accessed by brokers via the AgentCore Gateway to retrieve related knowledge. Rad calendar MCP device finds radiologists’ shifts and real-time schedules from healthcare scheduling programs for the radiologist availability sub-agent. Equally, the medical knowledge MCP device can discover the affected person’s historic medical knowledge for the affected person historical past synthesizer agent.
The next sub-agents are created.
- First is Rad project agent (rad_mapper) that matches radiologists primarily based on facility, web site, illness, subspecialty, affected person historic well being knowledge, medical notes, and different medical parameters, then categorizes them by precedence and solutions questions on radiologist particulars.
- Second is the Affected person historical past synthesizer agent (clinical_data_collector) that retrieves affected person medical historical past and identifies related historic data for radiologist project.
- Third is an Examination metadata synthesizer agent (metadata_finder) that extracts metadata from the present medical imaging examine to supply context (anatomy, notes, examination particulars) for radiologist project.
- Fourth is the Rad availability agent (shift_checker), which verifies radiologist availability and selects the very best out there radiologist from the filtered checklist by checking their schedules, present workload, and exceptions. The checklist is filtered by medical knowledge collector, metadata finder, and rad_mapper sub-agents.
By the AgentCore Gateway, brokers are supplied entry to PACS/Imaging API for querying examination metadata. AWS HealthImaging gives the cloud-native medical imaging repository, storing petabytes of DICOM photographs with sub-second retrieval speeds. It gives the examination metadata synthesizer agent with entry to imaging examine metadata together with affected person historical past, modality sort, physique elements examined, and urgency ranges.
The answer makes use of Amazon SageMaker AI to carry out real-time inference on machine studying fashions that detect acute, time-sensitive situations reminiscent of pulmonary embolism. These fashions analyze medical photographs saved in AWS HealthImaging and detect key findings that warrant speedy examination reprioritization. Inference outcomes are returned by way of the PACS/Imaging API to the brokers such because the examination prioritization agent, which dynamically adjusts worklist ordering primarily based on medical urgency.
On this resolution, AgentCore Observability is used to hint the total execution path when a question flows via the Clever worklist orchestrator and followers out to the Examination metadata, Scientific knowledge historical past, Rad mapper, Rad shift checker, and Dynamic guidelines brokers. Every agent invocation is captured as a hint with particular person spans, so when an examination project request takes longer than anticipated, it may possibly pinpoint whether or not the bottleneck was within the Scientific knowledge API name by way of MCP Gateway, a gradual reminiscence retrieval from AgentCore Reminiscence, or the LLM inference itself. The Trajectory view proven right here visualizes this end-to-end span chain for a single worklist question, making it easy to debug points like a Rad shift checker agent failing to retrieve calendar knowledge or the orchestrator routing to the improper sub-agent. These traces feed into Amazon CloudWatch dashboards that monitor per-agent latency, device invocation success charges, token consumption, and reminiscence learn/write patterns. This gives the operations crew the alerts they should tune agent efficiency and catch regressions earlier than they impression worklist throughput.

Cleanup

The code and directions to arrange and clear up this resolution can be found within the Clever radiology workflow optimization GitHub repo.

Conclusion

On this submit, we confirmed how transferring your radiology worklist administration from inflexible, rule-based programs to clever, agent-driven orchestration provides your group a sensible path to decreasing operational inefficiencies and defending your clinicians from burnout. The outcomes we have now walked via present that your workflows enhance not by including extra guidelines, however by deploying programs able to real reasoning, contextual judgment, and steady adaptation. You possibly can prolong this resolution additional to extend its worth. By analyzing examination quantity and complexity patterns, your brokers can establish workflow bottlenecks earlier than they turn out to be backlogs, enabling proactive scheduling changes reminiscent of bringing in extra radiologists early, exactly when and the place your knowledge exhibits demand will spike. If you find yourself prepared to maneuver ahead, begin by figuring out the highest-impact use instances in your individual setting. From there, set up strong integration patterns together with your present medical programs, and undertake a phased strategy that provides your resolution the time and knowledge it must be taught, refine, and constantly enhance.

Get began at the moment by contacting your AWS account consultant to debate a pilot implementation. To be taught extra, speak with your AWS account team.