The proposed paper suggests an Intelligent Decision Support System (IDSS) to support the management of

healthcare resources with the help of a Hybrid AI Architecture based on the fusion of Machine Learning (ML)

and Knowledge Graphs. The system also seeks to streamline resource allocation i.e. bed allocation, staff

scheduling and equipment allocation through predictive analytics and real time decisions. The key tools that we

use in this approach are Azure AI and Azure Knowledge Graph. The machine learning of Azure AI is employed

to create models that predict patient demand and resource needs, and the Azure Knowledge Graph organizes and

combines heterogeneous healthcare data, i.e., patient history, diagnosis and resource availability. Moreover,

Retrieval-Augmented Generation (RAG) is being introduced to offer live data based and predictive model

recommendations. This integrated solution guarantees scalable, adaptive, and transparent decision-making and

eventually, the efficiency of healthcare, lowers cost, and improves the quality of care using smart and data-driven

provision becomes more complicated, there is a need to adopt smart systems that will help achieve optimal

results in the distribution of resources in the form of beds, medical personnel, equipment, and medications. The

rather traditional approaches to healthcare resource management that frequently involve manual decision-

making and/or the use of fixed-point models can no longer sustain the dynamic nature of healthcare settings [1].

Consequently, the need to develop Intelligent Decision Support Systems (IDSS) that will automatize and

optimize the decision-making procedure through the application of the most recent technologies, such as

Machine Learning (ML) and Artificial Intelligence (AI), increases.

The purpose of IDSS in the healthcare sector is to assist medical practitioners and healthcare administrators

make informed decisions that enhance the efficiency of the use of the available resources and do not compromise

or deteriorate care quality [2]. The fact that such systems can process vast volumes of healthcare data and give

Demand predictive machine learning models such as predicting patient hospitalization, forecasting staffing, and

predicting the usage of medical equipment can be employed to predict demand of the hospital services [3]. These

predictions allow the administrators to make informed decisions in an attempted manner using the data, which

allows the healthcare entities to respond to the evolving demand and supply of resources more effectively.

Figure 1: Hybrid AI Revolutionizes Healthcare.

Quite to the contrary, knowledge graphs provide a semantically enriched model of healthcare data, such as

IDSS can also understand the mutual dependencies of numerous healthcare resources and stakeholders through

the assistance of a knowledge graph [4]. It allows making better decisions and being more aware of context, as

the system can build more complex relations and make recommendations based on the larger healthcare context.

The knowledge graph reasoning, as the combination of machine learning and knowledge graphs, will not only

predict the future resource requirements but also lead to the actionable insights of the proposed system. As an

example, it may be able to recommend the optimal strategies of resource allocation that factors in predicted

demand and available resources and justify why this has been done [6]. Moreover, to facilitate real-time decision

support, the system will be connected to Retrieval-Augmented Generation (RAG), which will provide the

healthcare professionals with context-sensitive suggestions grounded in live information, including the

availability of beds, staff, or equipment.

The current paper will discuss the design and implementation of this Hybrid AI Architecture to manage

healthcare resources and show how it would improve the effectiveness and responsiveness of healthcare systems

and address current drawbacks of the healthcare system in terms of scalability, adaptability, and interpretability

healthcare resources, this system can eventually revolutionize the process of managing healthcare resources in

order to enhance the healthcare provided to patients and business efficiency.

availability, and medical equipment utilization into a unified decision-making framework.

Furthermore, although knowledge graph technologies have been explored in healthcare information systems,

their integration with predictive machine learning models for operational decision support remains limited.

Existing approaches rarely combine predictive analytics with semantic reasoning and contextual retrieval

mechanisms, which are essential for handling the complex and dynamic nature of healthcare environments.

Another critical limitation in current research is the lack of hybrid architectures capable of supporting real-time

decision making in healthcare resource management. Most existing systems focus either on prediction or rule-

based decision support, without leveraging advanced frameworks such as retrieval-augmented generation (RAG)

to incorporate contextual knowledge and policy guidelines into decision processes. Consequently, there remains

a significant research gap in developing an integrated AI-driven framework that combines predictive modeling,

semantic knowledge representation, and contextual reasoning for effective healthcare resource management.

This study contributes to the growing body of research on intelligent healthcare management by proposing a

hybrid decision support framework that integrates machine learning, knowledge graph technology, and retrieval-

inflow, staff availability, and medical equipment utilization, the system enhances the capability of hospitals to

can represent relationships among healthcare entities, thereby improving interpretability and decision

transparency. Furthermore, the incorporation of retrieval-augmented mechanisms allows the system to utilize

contextual information such as clinical guidelines and institutional policies during decision making. The

There has been a strong investigation into the use of Intelligent Decision Support Systems (IDSS) and Decision

Support Systems (DSS) in different fields within the healthcare industry due to the necessity of creating

intelligent systems to assist in the management of healthcare resources. The systems are based on sophisticated

algorithms, machine learning, and knowledge representation techniques to enhance resource allocation, optimize

operations and lastly patient care [8]. The concept of artificial intelligence (AI) implementation in healthcare

management systems has been brought up in numerous publications and demonstrated the effective application

of artificial intelligence and issues that need to be addressed.

One of the most popular yet the oldest studies is resource demand forecasting using the application of Machine

Learning (ML). The main purpose of the initial research was to use machine learning frameworks such as

regression frameworks and decision trees to predict resource utilization in hospitals such as patient admission

rates, bed occupancy, and staffing levels. The article by Chien et al. (2019) [7] is one of the articles of particular

interest because the authors applied machine learning to predict patient admissions and optimize the bed

to predict intentional flow of patients and this helped the hospitals to plan during peak season and avoid

congestion. This, however, was not keeping up with dynamic changes in patient conditions or the ability to

accommodate complex patient related data which is also crucial in making holistic decisions.

To address such problems, researchers began to think about the use of knowledge graphs and semantic data

models to the health care decision support. The framework that has been suggested by Zhang et al. (2020) [9] is

a machine learning and knowledge graph combination in healthcare management. They used a knowledge graph

and applied it to structure and integrate healthcare data, such as medical history of patients, the presence of

This did enable the data to reason over by the system and provide a more informed recommendation especially

The application of ML and knowledge graph in healthcare decision support has also been expanded with the use

of reinforcement learning (RL) in making dynamic decisions. Sutton et al. (2021) suggested that RL algorithms

can be used to allocate resources in the healthcare system, with a particular emphasis on real-time decision-

making with the press of a button [11]. Their model will constantly learn and evolve according to the feedback

provided by the environment (e.g., patient outcomes, resource utilization patterns), which enables it to optimise

the utilisation of the resources in the long term.

The benefits of RL include the possibility to manage dynamic and multidimensional environments, i.e., when a

healthcare crisis (e.g., pandemics) strikes and the resource requirements may shift very quickly. Nevertheless,

RL models may also be computationally intensive, and thus may need large data volumes to train and may be

restrictive to real-time use in resource-constrained healthcare settings.

The incorporation of cloud-platforms is a large trend in healthcare decision support system development in recent

Cloud. The article by Radhakrishnan et al. (2022) employed the use of the Azure Intelligence to create an

intelligent staff scheduling system in hospitals that allows real-time staffing decisions [12]. Their system made

use of hospital management systems data to forecast their staffing requirements that lessened their wait times

et al. (2021) emphasize [13], explainable AI (XAI) is essential because it is necessary to make the decision

Scalability is another challenge of great importance. Most of the systems available are tailored to small-scale

applications or a particular department of a hospital, although they are not always scalable to multi-facility

healthcare systems. This limits their use in large healthcare systems which require management of resources in

more than one hospital or region [14].

To sum up, although the use of AI and machine learning in healthcare resource management has achieved

considerable progress, the combination of several technologies that may include machine learning, knowledge

graphs, and real-time decision-making is a nascent field. The mixture of the technologies, especially when it

limitations inherent to the data quality, scalability, interpretability, and real-time flexibility. With the ongoing

developments of healthcare systems, the purpose of intelligent, scalable, and transparent decision support

systems will be more than ever, and this study will become more applicable to the future developments in

Recent advancements in artificial intelligence have significantly influenced healthcare management, particularly

in improving operational efficiency and clinical decision support systems. Studies indicate that AI-driven models

have enhanced healthcare service quality by improving diagnostic accuracy, patient management, and resource

In recent years, knowledge graph technologies have emerged as an important component of intelligent healthcare

systems. Knowledge graphs enable the integration of heterogeneous healthcare data such as electronic health

records, clinical guidelines, and medical ontologies into structured networks that facilitate semantic reasoning

and explainable decision making (Shang et al., 2024; Yang et al., 2024). These graph-based models improve

interpretability and support clinical reasoning by representing complex relationships among healthcare entities

such as patients, diseases, treatments, and medical resources.

Several studies have also emphasized the importance of hybrid AI architectures that combine machine learning

techniques with knowledge representation frameworks to enhance clinical decision support systems. Integrating

symbolic reasoning with data-driven models enables healthcare systems to provide more reliable

recommendations and improves trust in AI-assisted medical decisions (Vidal et al., 2025; Jasthi, 2024).

retrieval and contextual reasoning in intelligent systems. RAG integrates external knowledge sources with

generative models to improve contextual understanding and provide more accurate responses for knowledge-

intensive applications such as healthcare decision support (Wang et al., 2025).

The creation of an Intelligent Decision Support System (IDSS) in the area of healthcare resource management

implies the combination of such advanced technologies as Machine Learning (ML), Knowledge Graphs, and

Cloud-based Infrastructure.

This is aimed at designing scalable, adaptive and real-time decision-making system which is optimized in

resource allocation like patient flow, bed management, staffing, and medical equipment and also improved the

quality of care and operational efficiency [15] . The research methodology is organized into a number of phases,

which include system design, data collection and preprocessing, model development, system integration, and

combination of two technologies: Machine Learning (ML) and Knowledge Graphs which improve decision-

making. Azure AI and Azure Knowledge Graph are chosen as the main tools of the platform because they are

scalable and flexible as well as have a wide range of functions that can assist in real-time data integration,

machine learning, and knowledge representation.

The ML models involve predicting future resources requirements with regard to past information and real-time

data of different hospital systems [16]. This involves forecasting the number of patients to be admitted, the

number of staff to be hired, and the number of medical equipment to be ordered.

Knowledge Graph component combines both structured and unstructured data to form a semantic model of the

relationships amongst various entities in the healthcare sector including patients, diseases, healthcare providers,

resources and medical conditions. This integration makes sure that the system has the ability of reasoning on the

data giving contextual decision support.

and medical devices [17]. The data contains the data on patient demographics, clinical history, admissions,

treatments, and resource utilization.

Preprocessing is an important process, considering the diversity and heterogeneity of healthcare data. This can

include methods like imputation to complete missing values, or data conversion to standardised formats and

formats to guarantee cross-compatibility of data sets [18]. Also, the data is standardized according to healthcare

requirements (e.g., ICD-10, SNOMED-CT, and HL7), which means that it can be successfully incorporated into

The IDSS lies in the fact that it creates machine learning models that forecast the various demands on healthcare

resources, including bed occupancy, staffing, and equipment utilization. Random Forest, XGBoost, and Neural

Networks are several of the supervised learning algorithms that are trained on historical data to make future

predictions on the demand of resources. The models are intended to perform classification (e.g. whether a patient

will be in need of critical care) and regression tasks (e.g. how many staff are needed on a particular day).

resources allocation continuously with incoming data. The RL agent is trained by his interactions with the