Manufacturer information and date of last update

Manufacturer information:
Tonic Easy Medical S.A.
Address: R. do Heroísmo, 281, Arm. 1, 4300-259 Porto, Portugal
Email: support@tonicapp.com
Website: www.tonicapp.io

Formal device name: AI Virtual Assistant
Trade name: Tonic Drugs
Basic UDI: ++G490121LZ
UDI-DI: +G4901210V
UDI-PI: +$$+71.2.1/16D20241128VC

Issuing organisation: HIBCC

Date of last update: 10.09.2025

Unique identifier: EN_V5.0_10092025

  1. What are the software and hardware requirements to access the Drugs (beta) model?

Compatible devices include iOS (Apple) and Android smartphones and tablets, as well as computers, smartphones and tablets that use modern web browsers such as Google Chrome, Mozilla Firefox, Safari and Microsoft Edge. You need an internet connection to use the tool, and we recommend keeping your operating system and browser up to date to ensure an optimal experience.

  1. Where can I access the Drugs (beta) model?

You can access it in the “Tonic AI” section of the Tonic platform.

  1. What is the Drugs (beta) model for?

This tool uses artificial intelligence (AI) and is designed to support you in caring for your patients by providing up-to-date, reliable information on medicines.

It has been trained by our team using the Summaries of Product Characteristics (SmPCs) of products registered with the MHRA (Medicines and Healthcare products Regulatory Agency). It is updated periodically. It is intended for use by doctors only and is designed to assist you, not replace your clinical judgement.

  1. What are the clinical benefits of the Drugs (beta) model?

The model optimises clinical care by helping you make more informed decisions, improving the safe use of medicines, providing personalised treatment recommendations and increasing the efficiency of clinical workflows.

  1. How does the Drugs (beta) model work? Can you give examples?

In Tonic, open the ‘Tonic AI’ section, select ‘Drugs’ from the available assistants using the buttons above the text input bar, and type your question.

A suggestion to get started: ‘Is clarithromycin indicated for the treatment of pneumonia?’ Once you receive the answer, you can ask follow-up questions because the tool retains context, e.g.: ‘What are the contraindications for this medicine?’.

  1. Can you give me some tips to get the highest-quality answers?

Answers are higher quality when you mention the medicine or molecule name in your question. If you want details about a specific brand, include it in your query. If you’re not fully satisfied with a response, you can regenerate the question or ask the tool to verify it and indicate the source document.

  1. What languages does the Drugs (beta) model use?

You can interact with the tool in several languages. However, we recommend asking your questions in English, as it is the primary language of the database used in the UK.

  1. Can I rely on the answers generated by the Drugs (beta) model?

The tool is continually evolving and being validated, but some answers may still be incomplete, imprecise or incorrect. It’s important that you review the documents provided with the answers by following the links, so you can verify the quality of each response. The information provided does not replace your clinical judgement nor the usual decision-making process for patient care. Always fact-check before making decisions or acting on the tool’s responses.

  1. What are the limitations of the Drugs (beta) model?

>The tool relies exclusively on information contained in Summaries of Product Characteristics (SmPCs) for products registered with the MHRA. If you have questions not covered in these documents (e.g., recommendations from medical societies), the tool will not be able to answer.

In addition, the tool cannot detect multiple medicines, molecules or conditions in a single query. It recognises and answers only the first entity mentioned, ignoring the others. To avoid this, ask separate questions for each medicine/molecule/condition rather than including multiple terms in one query. Finally, make sure all data you enter into the tool are accurate and complete, as incorrect inputs can lead to inappropriate recommendations.

  1. What are the contraindications for using the Drugs (beta) model?

Do not use it as the sole basis for therapeutic decisions; it is intended to support, not replace, your clinical judgement. Do not use the software in emergency situations where immediate clinical judgement is required.

  1. What residual risks have been identified in using the Drugs (beta) model?

Despite thorough design, verification and validation processes, the following residual risks persist:

Data privacy: although protection mechanisms exist, there remains a residual risk of unauthorised access to personally identifiable information.

Data integrity: there is a small risk of corruption or unauthorised modification of data, even with strict access controls.

System availability: external technical factors may temporarily affect access to the tool.

Bias in results: although mitigation strategies are implemented (e.g., randomised responses, traceability), a slight bias may persist and influence your interpretation.

User interpretation: the AI assistant is a support tool. Misinterpretation or excessive reliance without clinical verification may lead to inappropriate use.

Dependence on third-party models: the AI assistant partially depends on external model providers. Uncontrolled changes by these providers may affect reasoning or performance.

Recognition of multiple entities: the tool cannot detect multiple medicines, molecules or conditions in the same query. It recognises and answers only the first entity mentioned, ignoring the rest. To avoid this, ask separate questions for each medicine / molecule / condition rather than including multiple entities in a single question.

Various security and control measures have been implemented to reduce these risks, including data protection protocols, strict access controls and mechanisms to minimise bias. However, you should remain vigilant and verify the information obtained.

  1. Is the Drugs (beta) model up to date?

The tool may not reflect the most recent data if there have been recent changes to medicine information or updates to clinical guidance since the last software update. It’s important to stay alert to MHRA safety updates and notices.

  1. Is the Drugs (beta) model free?

At present, the tool is available free of charge to all our users, and we cover the cost of the service. Depending on usage volume, we may need to share costs in the future.

  1. Is there a message history?

The tool stores your last 20 messages. If you want to save any information, before leaving the chat you can use the “copy / paste” option by pressing and holding the message.

  1. How do I close or delete a chat?

To close the chat, simply return to the platform’s main screen. To delete it, click “Delete conversation and show examples” at the top of the chat window before you exit.

  1. What are the performance metrics of the Drugs (beta) model?

The tool has demonstrated excellent clinical performance across multiple dimensions, including completeness, accuracy, consistency and responsiveness.

Completeness: the model consistently retrieved comprehensive information on treatments and safety. In validation testing, completeness scores ranged from 92.96% to 99.67%, significantly outperforming general-purpose models such as GPT-4 and Med-PaLM (typical range: 60–98%).

Accuracy: the system achieved high accuracy across all validated use cases, with scores ranging from 90.14% to 100%, ensuring reliable retrieval of medicine names, active substances, SmPC sections and open clinical queries. This supports safe clinical decision-making by reducing the risk of misinformation.

Consistency: the model showed perfect consistency when answering similar queries across different SmPC versions and regional variants, with a Cronbach’s alpha of 1.00. This ensures reproducibility and clinician confidence in the results.

System performance: the assistant maintained low latency (8.24 seconds) and a 0% error rate under stress conditions, ensuring timely and reliable access to information.

Commercial bias: the model showed a Gini coefficient of 0.25, indicating an equitable distribution of medicine references across manufacturers.

  1. The Drugs (beta) model is slow – why?

During peak usage, the service may slow down. Please wait a moment and try again later.

  1. What should I do if the Drugs (beta) model fails?

If the tool is unavailable, make sure the device running the assistant is on and connected to the network. Close and reopen the platform software to restart the tool. Report the error immediately. You can do this by emailing support@tonicapp.com or via the ‘Support’ link at the top of the chat window. Your feedback is very important to us. As this is a beta version, we welcome your suggestions to improve the tool.

If the problem persists, consult another reliable source. You can look up Summaries of Product Characteristics (SmPCs) on the electronic Medicines Compendium (eMC): https://www.medicines.org.uk/emc.

  1. What is the privacy policy for the Drugs (beta) model?

Your interactions with the tool will not be shared with third parties — the generative AI technology used cannot learn from the data you share in the chat. In any case, please do not share sensitive third-party information (e.g., your patients’ personal data) during your conversations with the tool. For more information, see our privacy policy: https://tonicapp.io/privacy-policy/

  1. What are the terms of use for the Drugs (beta) model?

We do not permit the use of the tool to produce hateful, harassing or violent content, nor for illegal, fraudulent or misleading purposes. For more information, see our terms of use: https://tonicapp.io/terms-of-use/

  1. What should I do if a serious incident occurs related to the Drugs (beta) model?

If a serious incident occurs in connection with the use of the tool, inform the manufacturer and the competent authority – MHRA – immediately: https://www.gov.uk/mhra. Serious incidents may include situations that cause or could cause significant harm to the health of the user or the patient. This notification helps ensure the ongoing safety of the device and the implementation of corrective measures if needed.

  1. How can I access the electronic instructions for use for the Drugs (beta) model?

To access the tool’s instructions for use, follow these steps:
1- Go to our website: www.tonicapp.io
2- Select your preferred language in the navigation bar
3- Scroll to the bottom of the home page
4- Click ‘Tonic AI Instructions for Use’

If you have any difficulty, contact us at support@tonicapp.com for assistance.

To request paper instructions for use, email support@tonicapp.com with your request. The printed version of the instructions is not controlled. Please check for updates regularly by visiting the electronic instructions for use section on the website: www.tonicapp.io.

The instructions for use will be available for a period of 15 years after the last device has been placed on the market.