Intelligent medication management

Status: R&D project in progress

Pharma Flow

Medication management designed to stay ahead of problems.

Pharma Flow will support demand forecasting, inventory optimisation, contract monitoring and supplier communication.

Hospital pharmacist using conceptual Pharma Flow inventory, delivery and forecasting analyses
  • Demand forecasting
  • Inventory optimisation
  • Supplier communication

Planned project period: 1 March 2026–29 February 2028 · planned implementation of results: 1 March–31 August 2028

Planned scope The product is being developed within an R&D project. The described capabilities will be developed and verified through research, testing and pilots.

From recording events to anticipating needs

Medication management has to balance two opposing risks

A hospital pharmacy must protect medicine availability while avoiding excessive inventory that ties up funds and increases expiry risk. Changing ward demand, seasonality, uncertain deliveries, budget constraints and numerous contracts make forward-looking decisions difficult.

Too little inventory

A shortage may hinder or delay therapy and trigger time-consuming intervention purchasing.

Too much inventory

Excess stock ties up financial resources and increases the risk of medicine expiry.

Planned decision-support layer

Pharma Flow will complement the pharmacy system

Pharma Flow will use artificial intelligence, data analysis and federated learning to support demand forecasting, inventory optimisation, contract-performance monitoring and supplier communication. It will not replace a pharmacist or the software already used by the pharmacy.

  • Which medicine may run out before the next delivery?
  • Where may excess or expiry risk arise?
  • Is a supplier performing the contract as agreed?
  • Which order is delayed and requires action?
  • Is a proposed substitute consistent with the contract?
  • Where can a medicine be sourced in an emergency?

Planned operating model

How will Pharma Flow support the pharmacy?

The solution is intended to move medication management from reacting to recorded events towards earlier risk identification, an explained recommendation and an action approved by an authorised user.

  1. 01

    Use data from existing systems

    Historical consumption, inventory movements, stock levels, deliveries, orders, expiry dates, contract performance and selected hospitalisation or availability data may be analysed. The scope will be agreed with each hospital.

  2. 02

    Forecast demand

    Predictive models may consider seasonality, hospital characteristics, ward workloads, hospitalisation volumes, local epidemiological trends and unusual consumption changes.

  3. 03

    Recommend inventory levels

    The system may indicate shortage risk, excessive stock, expiry risk, a recommended order date and a suggested quantity for staff review.

  4. 04

    Monitor contract performance

    A planned natural-language-processing module will compare contract terms with actual deliveries and may flag delays, quantity or price differences and non-compliant substitutes.

  5. 05

    Organise supplier communication

    Standard correspondence and supplier replies may be connected with the relevant order, medicine, contract or case, while remaining under staff supervision.

  6. 06

    Support intervention orders

    If the regular supplier fails to perform, the system may help identify alternatives, send availability enquiries and compare responses. A user will approve the offer and order.

Planned dashboard

One view of forecasts, risks, orders and recommended actions

The main screen is intended to provide a concise view of the pharmacy situation and the items that may require attention. Each recommendation should explain what it concerns, the data behind it and the expected consequence.

Four pillars

A connected process rather than another isolated report

01

Demand forecasting

The system will forecast future medicine demand using historical data, hospital characteristics and time-dependent factors.

02

Inventory optimisation

Pharma Flow will help limit both shortage risk and inventory levels that exceed probable needs.

03

Contract monitoring

Planned NLP mechanisms will compare contract terms with delivery performance and flag potential discrepancies.

04

Communication and orders

The system will organise supplier contact and support sourcing a medicine when the standard supplier cannot deliver.

Federated learning

Hospitals retain source data; model updates support shared learning

Each partner hospital may train a local model using data that remains in its environment. The shared process will receive the parameters required to update a global model rather than patient source data or complete operational datasets.

Hospital ALocal dataLocal model
Hospital BLocal dataLocal model
Partner hospitalLocal dataLocal model
Model updates
Secure aggregationShared modelAnother version returned to participating hospitals
Conceptual flow: source data remains in hospital environments; only the information required for model updates is exchanged.
Research area requiring separate security analysis

Federated learning does not automatically eliminate every privacy risk. Architecture, security and the scope of exchanged information will be analysed and verified during the project.

Complementing existing software

Pharma Flow will not replace the pharmacy system

AMMS by Asseco, KS-ASW by Kamsoft, CGM Clininet and other HIS or pharmacy systems will remain responsible for source records, warehouse documents, settlements and operational data. Pharma Flow will create an additional analytical and recommendation layer.

Pharmacy system / HISSource data and operational records

Integration layerScope agreed with the hospital

Pharma FlowAnalysis · prediction · recommendation · communication

Forecasts · alerts · recommendationsInformation with justification

User decisionApproval by an authorised staff member
Planned information flow. The first integration is intended to use AMMS; subsequent integrations will draw on experience from that stage.

Benefits to be verified during implementation

One system, different perspectives within the hospital

Hospital pharmacist

  • Earlier warnings about possible shortages
  • Order-planning support and fewer manual analyses
  • Organised supplier communication
  • Explained recommendations with approval controls

Pharmacy manager

  • Concise view of inventory and contract performance
  • Control of shortage, excess and expiry risk
  • Supplier-performance information
  • Analysis of how recommendations are used

Management and finance

  • Better visibility of inventory value
  • Support for reducing funds tied up in stock
  • Analysis of delivery problems and their cost
  • Data supporting future contract negotiations

Procurement, administration and IT

  • Ordered contract and communication history
  • Support for monitoring dates and terms
  • Integration without replacing the pharmacy system
  • Control over the target deployment and data-flow model

Human oversight

The authorised user will remain at the centre of the process

Pharma Flow will support logistical, administrative and purchasing processes. It will not make clinical decisions, recommend patient treatment or autonomously submit an intervention order.

  • Change the planned order
  • Select an alternative supplier
  • Accept offer terms
  • Approve a substitute
  • Submit an intervention order
  • Act on a detected contract discrepancy

Planned deployment

From data analysis to a controlled production launch

After R&D, Pharma Flow is intended to be offered as SaaS or installed on premises. The final model will depend on security requirements, IT architecture, data-processing rules and hospital policy.

Model 01

SaaS

A cloud environment maintained by the supplier for hospitals that do not want to manage the application infrastructure themselves.

Model 02

On-premises

An installation in the hospital environment, with infrastructure and technical administration remaining under hospital control.

  1. 01

    Environment analysis

    Systems, pharmacy processes, available data and data quality.

  2. 02

    Data integration

    Mechanisms for obtaining data from the pharmacy system, HIS and other sources.

  3. 03

    Process configuration

    Rules for inventory, alerts, contracts, suppliers and action approval.

  4. 04

    Local model preparation

    Models adapted to the characteristics and data of the hospital.

  5. 05

    Pilot

    A limited observational and recommendation scope.

  6. 06

    Outcome assessment

    Comparison of forecasts, recommendations and detected issues with actual events.

  7. 07

    Production launch

    Gradual extension to further medicine groups and processes after results are accepted.

Project in progress

Planned project period through 29 February 2028

Pharma Flow is being developed under project FENG.01.01-IP.02-0638/25: “Development of the Pharma Flow intelligent agent using artificial intelligence and federated machine learning to optimise medication management and take over routine, time-consuming tasks performed by hospital pharmacy staff”.

The planned project period is 1 March 2026 – 29 February 2028. Implementation of the results is planned for the following six months: 1 March – 31 August 2028. At this stage, Pharma Flow is not a commercially ready product. The described capabilities are project assumptions that will undergo research, testing and validation.

Collaboration with hospitals

Partner hospitals may support process analysis, data-quality assessment, use-case consultation, model validation, prototype testing, usability assessment and pilots. Organisations interested in the project can already discuss future deployment scenarios during the R&D stage.

The future of medication management

Which pharmacy process still requires manual analysis of multiple reports, emails and telephone calls?

Let us discuss the processes that Pharma Flow is intended to improve and the scope of a possible pilot or research collaboration.