Un ospedale svizzero offre un algoritmo in grado di prevedere gli effetti collaterali avversi (ASE) per vaccini o farmaci in base alle caratteristiche fisiologiche e alla storia dei pazienti. Si cercano accordi di licenza con aziende che vogliono sfruttare il potenziale della previsione degli effetti collaterali più accurata e personalizzata per aumentare la sicurezza del paziente e / o guidare il paziente verso il trattamento più adatto.
Background
Adverse Side effects (ASE) for drugs or vaccines are a common occurrence. The risks of ASE described in the indication of drugs or vaccines is based on what was observed during clinical trials with a wide range of patients. However, these risks can drastically increase depending on the patient history and characteristics. Our algorithm leverages these data to allow personalized ASE prediction.
Technology Overview
The algorithm is based on input parameters including personal and medical characteristics of patients and gives refined and quantitative risks for a variety of side effects including both severe and non-severe reactions. The algorithm has successfully been applied so far to:
COVID vaccines: Pfizer, Moderna, AstraZeneca, Sputnik V and Sinopharm
Other vaccines: pneumococcal vaccine (Prevnar13), a flu vaccine (Fluarix), and a zoster vaccine (Shingrix).
Drugs: Revlimid and Avastin
Other drugs are subject to ongoing model developments.
Stage of development
TRL7 (algorithm can readily be used, although performance would benefit from training on bigger datasets). An initial version validated on 900 patients is available for assessment.
Applications and Benefits
- Guidance on the most appropriate and safest treatment according to a patient characteristic and history (or personalized medicine based on the safety of drugs and risk of ASE)
- Planning prophylactic measure to alleviate expected ASE (personalization of risk management planning for individual patients)
- Refining ASE indications for patients with certain conditions, characteristic or history
- Patient education through personalized factsheets: patients will be given explanatory information about possible ASE depending on the treatment options.
The University Hospital is looking for companies that are interested in licensing the algorithm and want to use it themselves for the personalized prediction of side effects of a drug or vaccination. The inventors are open to supporting the company technically.
- Accurate % risks of adverse side effects
The advantage can be understood with this real case scenario based on the data used to train the algorithm:
A subject was prescribed Revlimid for his Plasma Cell Myeloma condition. The subject died 90 days after the initial receiving of the Revlimid due to various ASE.
The subject experienced two major ASE due to Revlimid. There were Congestive Heart Failure (CHF) and Acute Renal Failure. The real-world chances of CHF and Renal Failure after receiving Revlimid are 1.7 and 2.2 percent, respectively. However, the personalised prediction approach reveals higher chances of these ASE for this individual.
The personalised prediction for CHF was that the patient has around an 8.2 percent chance of CHF (around 5 times the general risk).
The personalized prediction for Renal Failure was that the patient has around a 5.5 percent chance of Renal Failure (around 3.5 times the general risk).
The specific area of activity of the partner:
Digital health companies, pharmaceutical companies, clinical data and AI health companies
The tasks to be performed by the partner sought:
The industry partner wants to add this application to their portfolio to have an edge over the competition by offering personalized side-effect predictions for a drug or a vaccine.
Technological fields
-
Medical Technology / Biomedical Engineering
BIOLOGICAL SCIENCES / Medicine; Human Health
-
Clinical Research; Trials
BIOLOGICAL SCIENCES / Medicine; Human Health
-
Pharmaceutical Products / Drugs
BIOLOGICAL SCIENCES / Medicine; Human Health
-
Artificial Intelligence (AI)
ELECTRONICS; IT AND TELECOMMS / Information Processing & Systems; Workflow