FIELD OF THE INVENTION

The present invention generally relates to medication stock management. More particularly, the present invention relates to a system and a method for managing a medication stock of a healthcare facility using a combined deep learning and reinforcement learning.

BACKGROUND OF THE INVENTION

In pharmacy operation, management of inventory of medication stock is always a challenge for the healthcare delivery system. Without adequate pharmacy inventory management practices, healthcare facilities like hospitals and pharmacies run the risk of not being able to provide patients with the most appropriate medication when it is most needed. It also has a major impact on bottom line costs and profit margins. The management of inventory refers to all the medications and medical supplies used in daily operation of a healthcare facility. It comprises a large portion of the healthcare providers’ responsibilities where it is required to ensure an adequate stock of medications and supplies to serve the needs of the patients the facility serves.

A good medication stock management works towards preventing overstock items and understock items in the pharmacy. It is important to ensure medications are available when patients need them. Any overstock items may possess risks of medicines expiring, high inventory costs and limited storage space. Properly managing stock by using medications before they expire, and processing returns regularly can help keep medication costs down. The pharmacy with understock items may not be able to treat the expected case load of patients at the facilities concerned. The products that are regularly kept in stock are based on the needs of the pharmacy and its customers. While some rarely used, extremely expensive or cumbersome products may be ordered in as needed, efforts should be made to keep the medications used regularly in stock and available for use - not outdated or damaged. It is equally important in the medication stock order management to keep medication costs at a minimum. Traditionally, computerized management systems are used in the healthcare facilities to assist with the maintenance of inventory of medication stocks. One of the problems connected to the existing systems is the medicines needs for a given community in a geographical area to be covered are not well estimated. The existing estimation method also fails to factor in parameters such as the existing prescribing patterns and the procurement budget which led to poor costeffectiveness. Furthermore, the existing management systems are disadvantageous because they adopt analysis of previous data on medication stock order and estimation of medication stock order before placing a stock order to medication suppliers, that are labor-consuming and time-consuming.

By way of background, United States Patent Application Publication 2008/0270178 A1 discloses a system, method, and computer program product that are provided for an improved inventory management system. The system, according to the ‘178 publication, supports a medical service provider by interfacing an inventory management application with a database of electronic medical records. In one embodiment of the ‘178 publication, a single graphical user interface interfaces the inventory management application with the database of electronic medical records, thus providing a simplified, transparent solution as compared to the separate, independent, and non-interfaced systems currently available. By interfacing the inventory management application with the database of electronic medical records, inventory accuracy may be increased and processing times may be decreased.

It would, therefore, be advantageous to provide a solution that would overcome the deficiencies and shortcomings of prior art by way of providing a system and a method for managing a medication stock of a healthcare facility using a combined deep learning and reinforcement learning. Although there are systems and methods for the same in the prior art, for many practical purposes, there is still considerable room for improvement.

SUMMARY OF THE INVENTION

The following presents a simplified summary of the invention in order to provide a basic understanding of some aspects of the invention. This summary is not an extensive overview of the invention. Its sole purpose is to present some concepts of the invention in a simplified form as a prelude to the more detailed description that is presented later.

Accordingly, the present invention provides a system for managing a medication stock of a healthcare facility.

The system of the present invention may be characterized by a data harmonization module configured for processing, using a standardizing engine based on natural language processing, free-text clinical data obtainable from one or more healthcare facilities to generate harmonized and codified data, wherein the harmonized and codified data includes demographic data, morbidity data, services data, and consumption data harmonized and codified thereof, wherein the data harmonization module comprises a web crawler unit for obtaining crawler data associated with the said medication stock from world wide web; a morbidity pattern analyzing module configured for determining, using a morbidity pattern machine learning model, a first quantity estimation of stock order, wherein the morbidity pattern machine learning model is fed with a set of trend data generated upon analysis of the crawler data, the harmonized and codified demographic data, the harmonized and codified morbidity data, the harmonized and codified services data and a medication inventory associated thereof; a consumption analyzing module configured for determining, using a consumption machine learning model, a second quantity estimation of stock order, wherein the consumption machine learning model is fed with consumption trend data generated upon analysis of the crawler data, the harmonized and codified consumption data and the medication inventory thereof; a stock order recommendation module configured for providing, using a stock order machine learning model, a stock order recommendation based on a comparison conducted on each medication stock in respect of a budget-based stock order quantity, the first quantity estimation and the second quantity estimation against a recommendation threshold, wherein the recommendation threshold includes a stock order quantity threshold and a remaining procurement budget for a procurement period; and a stock order evaluation module configured for performing a cost-effectiveness evaluation for use in the said stock order recommendation module by way of processing the first quantity estimation, the second quantity estimation and computed costs associated thereof against a cost threshold. It is preferred that the morbidity pattern machine learning model, the consumption machine learning model and the stock order machine learning model utilize a combined deep learning and reinforcement learning framework. Preferably, the set of trend data includes demographic trend data, morbidity trend data and services trend data.

Preferably, the medication inventory comprises medication stock records including medication stock in hand and expiry data thereof.

Preferably, the demographic data includes historical demographic data and current demographic data.

Preferably, the morbidity data includes historical morbidity data and current morbidity data.

Preferably, the services data includes historical services data and current services data.

Preferably, the consumption data includes historical consumption data and current consumption data.

Preferably, the budget-based stock order quantity is determined based on the remaining procurement budget and unit price data associated with the medication stock thereof.

In accordance with another aspect of the present invention, there is provided a method for managing a medication stock of a healthcare facility.

The method may be characterized by the steps of processing, using a standardizing engine based on natural language processing, free-text clinical data obtainable from one or more healthcare facilities to generate harmonized and codified data, wherein the harmonized and codified data includes demographic data, morbidity data, services data, and consumption data harmonized and codified thereof, wherein the step of processing includes the step of obtaining crawler data associated with the said medication stock from world wide web; determining, using a morbidity pattern machine learning model, a first quantity estimation of stock order, including generating a set of trend data for feeding to the morbidity pattern machine learning model by way of analyzing the crawler data, the harmonized and codified demographic data, the harmonized and codified morbidity data, the harmonized and codified services data and a medication inventory associated thereof; determining, using a consumption machine learning model, a second quantity estimation of stock order, including generating a consumption trend data for feeding to the consumption machine learning model by way of analyzing the crawler data, the harmonized and codified consumption data and the medication inventory thereof; providing, using a stock order machine learning model, a stock order recommendation based on a comparison conducted on each medication stock in respect of a budget-based stock order quantity, the first quantity estimation and the second quantity estimation against a recommendation threshold, wherein the recommendation threshold includes a stock order quantity threshold and a remaining procurement budget for a procurement period; and performing a costeffectiveness evaluation for use in the stock order recommendation by way of processing the first quantity estimation, the second quantity estimation and computed costs associated thereof against a cost threshold. It is preferred that the morbidity pattern machine learning model, the consumption machine learning model and the stock order machine learning model utilize a combined deep learning and reinforcement learning framework.

The foregoing and other objects, features, aspects and advantages of the present invention will become better understood from a careful reading of a detailed description provided herein below with appropriate reference to the accompanying drawings.

BRIEF DESCRIPTION OF THE DRAWINGS

A more complete appreciation of the invention and many of the attendant advantages thereof will be readily as the same becomes better understood by reference to the following detailed description when considered in connection with the accompanying drawings, wherein:

Figure 1 is a schematic diagram of a system for managing a medication stock of a healthcare facility according to one embodiment of the present invention;

Figure 2 is an overall architecture of the system of Figure 1 according to one embodiment of the present invention; Figure 3 is a flow diagram of a method for managing a medication stock of a healthcare facility according to one embodiment of the present invention;

Figure 4 is a flow diagram of a data harmonization module representing the step of processing, using a standardizing engine based on natural language processing, free-text clinical data obtainable from one or more healthcare facilities to generate harmonized and codified data as stated in the method of Figure 3 according to one embodiment of the present invention;

Figure 5 is a flow diagram of a morbidity pattern analyzing module representing the step of determining, using a morbidity pattern machine learning model, a first quantity estimation of stock order as stated in the method of Figure 3 according to one embodiment of the present invention;

Figure 6 is a flow diagram of a consumption analyzing module representing the step of determining, using a consumption machine learning model, a second quantity estimation of stock order as stated in the method of Figure 3 according to one embodiment of the present invention;

Figure 7 is a flow diagram of a stock order recommendation module representing the step of providing, using a stock order machine learning model, a stock order recommendation based on a comparison conducted on each medication stock in respect of a budget-based stock order quantity, the first quantity estimation and the second quantity estimation against a recommendation threshold as stated in the method of Figure 3 according to one embodiment of the present invention;

Figure 8 is a flow diagram of a stock order evaluation module representing the step of performing a cost-effectiveness evaluation for use in the stock order recommendation by way of processing the first quantity estimation, the second quantity estimation and computed costs associated thereof against a cost threshold as stated in the method of Figure 3 according to one embodiment of the present invention;

Figure 9 shows an illustration of the flow diagram in Figure 4 pertaining to the data harmonization module according to one embodiment of the present invention; Figure 10 shows an illustration of the flow diagram of the morbidity pattern machine learning model of the morbidity pattern analyzing module in Figure 5, and the consumption machine learning model of the consumption analyzing module in Figure 6 according to one embodiment of the present invention;

Figure 11 shows an illustration of the flow diagram of the stock order machine learning model of the stock order recommendation module in Figure 7 according to one embodiment of the present invention;

Figure 12 shows an illustration of the analyses of historical demographic trends, historical morbidity trends and historical services trends employed at the morbidity pattern analyzing module in Figure 5 according to one embodiment of the present invention; and

Figure 13 shows an illustration of the analyses of historical consumption trends employed at the consumption analyzing module in Figure 6 according to one embodiment of the present invention.

It is noted that the drawings may not be to scale. The drawings are intended to depict only typical aspects of the invention, and therefore should not be considered as limiting the scope of the invention.

DETAILED DESCRIPTION OF THE INVENTION

The present invention discloses a system and a method for providing for managing a medication stock of a healthcare facility using a combined deep learning and reinforcement learning. The present invention essentially provides predictive analytics of next medication stock order for pharmacies within healthcare facilities that leverage on morbidity patterns and prediction. The system of the present invention improves the management of inventory of medication stock by rendering predictions of the next medication stock order based on data harmonization, i.e. codified data, and stock order prediction machine learning model. It will be evident that the present invention is in no manner limited to the medication stock management but may be applied to other stock managements and inventory systems. According to one preferred embodiment, the system of the present invention preferably comprises a data harmonization module 100, a morbidity pattern analyzing module 200, a consumption analyzing module 300, a stock order recommendation module 400 and a stock order evaluation module 500 as schematically shown in Figures 1 and 2. In general, the system of the present invention processes inputs such as procurement period and procurement budget in order to identify or determine a medication stock order for the healthcare facility.

Figure 1 schematically shows the system of the present invention which comprises the data harmonization module 100, the morbidity pattern analyzing module 200, the consumption analyzing module 300, the stock order recommendation module 400 and the stock order evaluation module 500. The system of the present invention is connected to various databases like rules database, harmonized data base, medication inventory database, consumption prediction database, morbidity prediction database, estimated stock order quantity database, and cost-effectiveness evaluation/specification database. The system is also connected to data inputs such as consumption machine learning model input, morbidity pattern machine learning model input, and stock order machine learning model input as well as procurement period and budget input.

With reference to Figure 2, the data harmonization module 100 is preferably connected to the morbidity pattern analyzing module 200 and the consumption analyzing module 300. The morbidity pattern analyzing module 200 and the consumption analyzing module 300 are connected to the stock order recommendation module 400. The stock order recommendation module 400 is further connected to the stock order evaluation module 500.

The data harmonization module 100 which is preferably a data processing engine may be configured for processing free-text clinical data using a standardizing engine based on natural language processing (NLP) to generate harmonized and codified data. The free-text clinical data is preferably obtainable from one or more healthcare facilities. The data harmonization module 100 preferably harmonizes and codifies the free-text clinical data. The data harmonization module 100 has a database connected thereto for storing the harmonized and codified data therein. In one embodiment of the present invention, the harmonized and codified data includes, but not limited to, demographic data, morbidity data, services data, and consumption data which are harmonized and codified accordingly. The demographic data includes historical demographic data and current demographic data. The morbidity data includes historical morbidity data and current morbidity data. The services data includes historical services data and current services data. The consumption data includes historical consumption data and current consumption data.

The standardizing engine of the data harmonization module 100, for instance, may adopt a standard coding scheme such as SNOMED (Systemized Nomenclature of Human and Veterinary Medicine) or other systemically organized computer processable collection of terminology covering areas of concerned.

The data harmonization module 100 comprises a web crawler unit. The web crawler unit is preferably configured for obtaining or fetching and analyzing crawler data associated with the said medication stock from world wide web, e.g. web pages. It is preferred that the web crawler unit extracts web page information. The crawler data retrieved thereof may be subject to the standardizing engine for harmonization and codification.

The morbidity pattern analyzing module 200 which is preferably an engine may be configured for determining, using a morbidity pattern machine learning model, a first quantity estimation of stock order. In one preferred embodiment, the morbidity pattern machine learning model utilizes a combined deep learning and reinforcement learning framework. It is continuously trained and improved over the time using the said combined deep learning and reinforcement learning framework.

The morbidity pattern machine learning model is preferably fed with a set of trend data generated upon analysis of the crawler data, the harmonized and codified demographic data, the harmonized and codified morbidity data, the harmonized and codified services data and a medication inventory associated thereof. The morbidity pattern analyzing module 200 preferably analyzes, summarizes and predicts the change in program targets based on the set of trend data which includes the demographic trend data, the morbidity trend data and the services trend data. In one embodiment, the medication inventory preferably comprises medication stock records including medication stock in hand and expiry data thereof.

The morbidity pattern analyzing module 200 is connected to databases including a rule database and a morbidity prediction database.

The consumption analyzing module 300 which is preferably an engine may be configured for determining, using a consumption machine learning model, a second quantity estimation of stock order. In one preferred embodiment, the consumption machine learning model utilizes a combined deep learning and reinforcement learning framework. It is continuously trained and improved over the time using the said combined deep learning and reinforcement framework.

The consumption analyzing module 300 is preferably fed with consumption trend data generated upon analysis of the crawler data, the harmonized and codified consumption data and the medication inventory thereof. The consumption analyzing module 300 preferably analyzes, summarizes and predicts the change in consumption based on the historical consumption trend data.

The consumption analyzing module 300 is connected to databases including a rule database and a consumption prediction database.

The stock order recommendation module 400 which is preferably an engine may be configured for providing, using a stock order machine learning model, a stock order recommendation. The stock order recommendation is determined based on a comparison conducted on each medication stock in respect of a budgetbased stock order quantity, the first quantity estimation and the second quantity estimation against a recommendation threshold. The budget-based stock order quantity is preferably determined based on the remaining procurement budget and unit price data associated with the medication stock thereof.

In one embodiment, the recommendation threshold includes a stock order quantity threshold and a remaining procurement budget for a procurement period. Essentially, the stock order recommendation module 400 predicts the next stock order quantity within the procurement period and the procurement budget. According to one exemplary embodiment of the present invention, the stock order recommendation module 400 adopts the following tables.

Table 1 : Medication Stock with Respective Stock Order Quantity Threshold.

Table 2: Calculation of Estimated Total Cost Per Product based on Stock Order Recommendation and Budget-based Stock Order Quantity.

Note: o Estimated Product Cost = (Stock Order Quantity Threshold - Medication Stock In Hand) x Unit Price o Budget-based Stock Order Quantity = (Remaining Procurement Budget - Shipping Cost) / Unit Price o Stock Order Recommendation = If Budget-based Stock Order Quantity + Medication Stock In Hand < Stock Order Quantity Threshold Choose Budget-based Stock Order Quantity, else MIN(Budget-based Stock Order Quantity, Stock Order Quantity Threshold - Medication Stock In Hand) o Estimated Total Cost Per Product = (Stock Order Recommendation x Unit Price) + Shipping Cost o Estimated Total Cost = SUM(Estimated Total Cost Per Product)

In one preferred embodiment, the stock order machine learning model of the stock order recommendation module 400 utilizes a combined deep learning and reinforcement learning framework. It is continuously trained and improved over the time using the said combined deep learning and reinforcement learning framework.

The stock order evaluation module 500 which is preferably an engine may be configured for performing a cost-effectiveness evaluation for use in the said stock order recommendation module 500. The cost-effectiveness evaluation is conducted by way of processing the first quantity estimation, the second quantity estimation and computed costs associated thereof against a cost threshold set therefor. The stock order recommendation preferably comprises the medication stock order which has been subject to the cost-effectiveness evaluation thereof.

The stock order evaluation module 500 is connected to databases including an estimated total cost database and a cost-effective specification database.

An overall relationship between the modules of the system in the present invention is described in Figure 2. In particular, the data harmonization module 100, upon retrieval of the harmonized and codified data from the database connected thereof, checks for consumption data.

If the consumption data associated with a concerned medication stock is not available, then the data harmonization module 100 will feed the retrieved harmonized and codified data to the morbidity pattern analyzing module 200. Based on the set of trend data generated upon analysis of the crawler data, the harmonized and codified demographic data, the harmonized and codified morbidity data, the harmonized and codified services data and the medication inventory thereof, the morbidity pattern analyzing module 200 determines the first quantity estimation of stock order using the morbidity pattern machine learning model thereof. The computed first quantity estimation of stock order will next be fed to the stock order recommendation module 400.

If the consumption data associated with a concerned medication stock is available, then the data harmonization module 100 will feed the consumption data to the consumption analyzing module 300. Based on the consumption trend data generated upon analysis of the crawler data, the harmonized and codified consumption data and the medication inventory thereof, the consumption analyzing module 300 determines the second quantity estimation of stock order using the consumption machine learning model thereof. The computed second quantity estimation of stock order will next be fed to the stock order recommendation module 400.

Once received, the stock order recommendation module 400 will subject the computed first quantity estimation of stock order or the computed second quantity estimation of stock order or a combination thereof to the stock order machine learning model. The stock order recommendation module 400 compares the budget-based stock order quantity, the first quantity estimation and/or the second quantity estimation against the recommendation threshold to produce the stock order recommendation.

The stock order recommendation will be subsequently fed to the stock order evaluation module 500. The stock order evaluation module 500 performs the costeffectiveness evaluation onto the said stock order recommendation. Once completed, it is preferred that the stock order evaluation module 500 generates a proposed medication stock order for each medication stock. As with the method of the present invention, Figure 3 provides a summarized flow diagram depicting the steps involved, as explained in the preceding paragraphs in connection with the system thereof. For purposes of clarity in explanation and understanding, the method of the present invention will be described in the following section.

The method preferably begins with step 600 of processing free-text clinical data obtainable from one or more healthcare facilities to generate harmonized and codified data using the standardizing engine based on natural language processing. It is preferred that step 600 includes step 600a of obtaining crawler data associated with the medication stock from the world wide web.

In one embodiment of the data harmonization module 100, with reference to Figure 4, step 600 preferably includes the following: a) receiving the free-text clinical data from an electronic health record (containing raw data); b) converting the said data to a lower-case character or letter code; c) subjecting the converted data to a pre-processing function including tokenization; d) subjecting the converted data to another pre-processing function including modifiers such as negation to interpretation of the same; e) subjecting the converted data to codification and tagging that is facilitated by a standard coding scheme such as SNOMED and a reference set (refer RefSet); and f) storing the codified records resulting thereof in the harmonized data base.

Figure 9 illustrates the flow diagram in Figure 4 pertaining to the data harmonization module 100 according to one embodiment of the present invention. In the illustration, there are three healthcare facilities, e.g. hospitals, each provides the free-text clinical data (see “File A", “File B” and “File C” in Figure 9). The free- text clinical data retrieved from these hospitals will be subject to a network protocol such as Secure File Transfer Protocol (SFTP) that provides file access, file transfer and file management over a reliable data stream to an ETL (extract, transform, load) platform. The ETL platform preferably extracts the said free-text clinical data and transforms the same to a suitable format (usually to fit a model of data warehouse) for use in a distributed computing application such as Hadoop being the data warehouse. During runtime of the standard coding scheme (e.g. SNOMED), the data harmonization module 100 generates the harmonized and codified data (see the codified files and the raw files in cycle) and store the same in the harmonized data base connected thereto.

It is subsequently followed by step 601 of determining the first quantity estimation of stock order using the morbidity pattern machine learning model. It is preferred that step 601 includes step 601a of generating the set of trend data for feeding to the morbidity pattern machine learning model by way of analyzing the crawler data, the harmonized and codified demographic data, the harmonized and codified morbidity data, the harmonized and codified services data and a medication inventory associated thereof.

In one embodiment of the morbidity pattern analyzing module 200, with reference to Figure 5, step 601 preferably includes the following: a) querying demographic, morbidity and services data from the harmonized data base, and the medication inventory database; b) analyzing historical demographic trends from a historical demographic database which shall provide historical demographic data, the latest demographic data and rules associated thereof; c) analyzing historical morbidity trends from a historical morbidity database which shall provide historical morbidity data, the latest morbidity data and rules associated thereof; d) analyzing historical services trends from a historical services database which shall provide historical services data, the latest services data and rules associated thereof; e) summarizing and predicting a change in program targets or coverage using the morbidity pattern machine learning model and the rules thereof; f) storing the change in the program target reflecting the morbidity predicted thereof in the morbidity prediction database; g) computing and storing an estimated stock order quantity; and h) establishing a first quantity estimation of stock order. The flow diagram of the morbidity pattern machine learning model of the morbidity pattern analyzing module 200 in Figure 5 is illustrated in Figure 10. The morbidity pattern machine learning model of the morbidity pattern analyzing module 200 is fed with and trained by the historical demographic trends, the historical morbidity trends, and the historical services trends analyzed thereof. Figure 12 provides examples of the historical demographic trends, the historical morbidity trends, and the historical services trends. The historical demographic trends comprise historical data including, but not limited to, divorce rate, current pandemic (e.g. COVID-19, SARS), eco-friendliness, live birth, unemployment rate, stress level and life style. The historical morbidity trends comprise historical data including, but not limited to, suicide rate, death rate, psychology impact, obesity rate, demotivation, panic attack rate, health status and hypertension rate. The historical services trends comprise historical data including, but not limited to, extend or limit of healthcare facilities, number of beds, staffs and equipment, essential services, stock out, supply rate and demand rate.

The morbidity pattern analyzing module 200 in turn subjects these trends to a combination of reinforcement learning (RL) and deep learning (DL) which learn autonomously. The RL has a function approximation by a deep neural network, and is preferably defined by three components, namely states, actions and rewards. Briefly, states are a representation of the current world or environment of the task. Actions are something an RL agent can do to change these states. Rewards are the utility the agent receives for performing the “right” actions. The RL in the present invention learns dynamically by way of adjusting the actions using a continuous feedback (e.g. trial and error method) in order to optimize the rewards. The DL preferably learns from a training set (e.g. existing knowledge) and then, applies its learning to a new data set.

Following that, the method initiates step 602 of determining the second quantity estimation of stock order using the consumption machine learning model. It is preferred that step 602 includes step 602a of generating the consumption trend data for feeding to the consumption machine learning model by way of analyzing the crawler data, the harmonized and codified consumption data and the medication inventory thereof.

In one embodiment of the consumption analyzing module 300, with reference to Figure 6, step 602 preferably includes the following: a) querying consumption data from the harmonized data base, and the medication inventory database; b) analyzing historical consumption trends from a historical consumption database which shall provide historical consumption data, the latest consumption data and rules associated thereof; c) summarizing and predicting a change in consumption using the consumption machine learning model and associated thereof; d) storing the change in the consumption reflecting the consumption predicted thereof in the consumption prediction database; e) computing and storing an estimated stock order quantity in the estimated stock order quantity database; and f) establishing a second quantity estimation of stock order and storing the same in a second quantity estimation of stock order.

The flow diagram of the consumption machine learning model of the consumption analyzing module 300 in Figure 6 is illustrated in Figure 10. The consumption machine learning model of the consumption analyzing module 300 is fed with and trained by the consumption trends analyzed thereof. Figure 13 provides examples of the historical consumption trends. The historical consumption trends comprise historical data including, but not limited to, manufacturing issues, allergy, shortage of ingredients, risk level, drug abuse, malpractice, redundant prescription, counterfeit medicine, stop or cease of production, over prescription, pendency of approval from authorities (e.g. health ministry), alternative medication, stock out, legal issues, import issues, banned medicine, limited medication supply, poverty, increment in medication doses, and medication regime change.

The consumption analyzing module 300 in turn subjects the consumption trends to a combination of RL and DL. As described in the preceding paragraphs, the RL has a function approximation by a deep neural network, and is preferably defined by three components, namely states, actions and rewards. Briefly, states are a representation of the current world or environment of the task. Actions are something an RL agent can do to change these states. Rewards are the utility the agent receives for performing the “right" actions. The RL in the present invention learns dynamically by way of adjusting the actions using a continuous feedback (e.g. trial and error method) in order to optimize the rewards. The DL preferably learns from a training set (e.g. existing knowledge) and then, applies its learning to a new data set.

Thereafter, the method executes step 603 of providing the stock order recommendation using the stock order machine learning model based on the comparison conducted on each medication stock in respect of a budget-based stock order quantity, the first quantity estimation and the second quantity estimation against the recommendation threshold.

In one embodiment of the stock order recommendation module 400, with reference to Figure 7, step 603 preferably includes the following: a) retrieving a medication stock in hand from the medication inventory; b) computing expiry data for the said medication stock, wherein the computation includes those nearly expired batch of the medication stock thereof; c) computing a ready medication stock after lead time from the identified nearly expired batch of the medication stock; and d) comparing the ready medication stock against a minimum stock level threshold, wherein if the ready medication stock is less than the said minimum stock level threshold, then increasing the number of the ready medication stock which has the highest demand; wherein if the ready medication stock is equal to or greater than the said minimum stock level threshold, then retrieving the predicted stock order quantity, particularly the first quantity estimation of stock order and the second quantity estimation of stock order respectively from the morbidity pattern analyzing module 200 and the consumption analyzing module 300 via the respective databases; computing an estimated total cost; and comparing the said estimated total cost against a procurement budget threshold, wherein if the estimated total cost is greater than the said procurement budget threshold, then reducing the number of the ready medication stock which is deemed slow or non-moving stock; wherein if the estimated total cost is equal to or lower than the said procurement budget threshold, then storing the estimated or predicted total cost thereof in; and updating the estimated stock order quantity in the estimated stock order quantity database.

An illustration of the flow diagram of the stock order machine learning model of the stock order recommendation module 400 in Figure 7 is shown in Figure 11. The stock order machine learning model of the stock order recommendation module 400 is fed with and trained by data including the first quantity estimation and the second quantity estimation from the estimated stock order quantity database including the first quantity estimation of stock order database and the second quantity estimation of stock order database, the medication inventory from the medication inventory database, and the procurement budget for a particular procurement period. The stock order recommendation module 400 in turn subjects the data thereof to a combination of RL and DL. As described in the preceding paragraphs, the RL has a function approximation by a deep neural network, and is preferably defined by three components, namely states, actions and rewards. Briefly, states are a representation of the current world or environment of the task. Actions are something an RL agent can do to change these states. Rewards are the utility the agent receives for performing the “right” actions. The RL in the present invention learns dynamically by way of adjusting the actions using a continuous feedback (e.g. trial and error method) in order to optimize the rewards. The DL preferably learns from a training set (e.g. existing knowledge) and then, applies its learning to a new data set.

Once the stock order recommendation is produced, the method initiates step 604 of performing the cost-effectiveness evaluation for use in the stock order recommendation by way of processing the first quantity estimation, the second quantity estimation and computed costs associated thereof against the cost threshold. In one embodiment of the stock order evaluation module 500, with reference to Figure 8, step 604 preferably includes the following: a) retrieving the first quantity estimation and the second quantity estimation from the estimated stock order quantity database including the first quantity estimation of stock order database and the second quantity estimation of stock order database; b) retrieving the estimated total cost from the estimated total cost database; c) retrieving a cost-effective specification from the cost-effective specification database; d) performing the cost-effectiveness evaluation and comparing the same against a cost acceptable threshold, wherein if the cost-effectiveness evaluation meets the said cost acceptable threshold, then outputting a medication stock order; wherein if the cost-effectiveness evaluation fails to meet the said cost acceptable threshold, then re-initiating step 603 of providing the stock order recommendation at the stock order recommendation module 400.

Although an overview of the inventive subject matter has been described with reference to specific example embodiments, various modifications and changes may be made to these embodiments without departing from the broader scope of embodiments of the present disclosure. Such embodiments of the inventive subject matter may be referred to herein, individually or collectively, by the term “invention” merely for convenience and without intending to voluntarily limit the scope of this application to any single disclosure or inventive concept if more than one is, in fact, disclosed.

The embodiments illustrated herein are described in sufficient detail to enable those skilled in the art to practice the teachings disclosed. Other embodiments may be used and derived therefrom, such that structural and logical substitutions and changes may be made without departing from the scope of this disclosure. The Detailed Description, therefore, is not to be taken in a limiting sense, and the scope of various embodiments is defined only by the appended claims, along with the full range of equivalents to which such claims are entitled. As used herein, the term “or” may be construed in either an inclusive or exclusive sense. Moreover, plural instances may be provided for resources, operations, or structures described herein as a single instance. Additionally, boundaries between various resources, operations, modules, engines, and data stores are somewhat arbitrary, and particular operations are illustrated in a context of specific illustrative configurations. Other allocations of functionality are envisioned and may fall within a scope of various embodiments of the present disclosure. In general, structures and functionality presented as separate resources in the example configurations may be implemented as a combined structure or resource. Similarly, structures and functionality presented as a single resource may be implemented as separate resources. These and other variations, modifications, additions, and improvements fall within a scope of embodiments of the present disclosure as represented by the appended claims. The specification and drawings are, accordingly, to be regarded in an illustrative rather than a restrictive sense.

The foregoing description, for the purpose of explanation, has been described with reference to specific example embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the possible example embodiments to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings. The example embodiments were chosen and described in order to best explain the principles involved and their practical applications, to thereby enable others skilled in the art to best utilize the various example embodiments with various modifications as are suited to the particular use contemplated.

It will also be understood that, although the terms “first,” “second,” and so forth may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first contact could be termed a second contact, and, similarly, a second contact could be termed a first contact, without departing from the scope of the present example embodiments. The first contact and the second contact are both contacts, but they are not the same contact.

The terminology used in the description of the example embodiments herein is for the purpose of describing particular example embodiments only and is not intended to be limiting. As used in the description of the example embodiments and the appended examples, the singular forms “a," “an,” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will also be understood that the term “and/or” as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items. It will be further understood that the terms “comprises” and/or “comprising,” when used in this specification, specify the presence of stated features, integers, steps, operations, elements, and/or components, but do not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and/or groups thereof.

As used herein, the term “if may be construed to mean “when” or “upon” or “in response to determining” or “in response to detecting,” depending on the context. Similarly, the phrase “if it is determined" or “if [a stated condition or event] is detected” may be construed to mean “upon determining” or “in response to determining” or “upon detecting [the stated condition or event]” or “in response to detecting [the stated condition or event],” depending on the context.