Bioinformatics in mission success FDA-CDRH

Document Sample
Bioinformatics in mission success FDA-CDRH Powered By Docstoc
					U. S. Department of                                                 Center for Devices and
Health and Human Services                                           Radiological Health




                    Bioinformatics in mission
                    success FDA-CDRH

                                   Brian Fitzgerald
                            Brian.fitzgerald@fda.hhs.gov
                                    (301) 796-2579

                                  Bioinformatics technology forum
    Regulatory Mission                                     Center for Devices and
                                                           Radiological Health


 Safe and effective medical devices
 Effective is mostly a clinical decision
       Does it do what it says it will (evidence based)
       Is it really a “pulse oximeter” (or something else too)



 Safe is an engineering approximation
       Never 100% safe or else too expensive
       Is „safe enough‟ actually „good enough‟
       So how to;
            Establish that its good enough
            Communicate to the regulator that its good enough
Traditional engineering practice         Center for Devices and
                                         Radiological Health


   Train
   Monitor
   Review
                 }      Professional discipline



   Use the “currently acknowledged state of the
    art”.
        Standards
        Regulation
Modern reality                                Center for Devices and
                                              Radiological Health


 Development costs are very high
      Single exemplar cost
      Expertise necessary not easily available

 Complexity is astonishing
      Analysing   the software for defects
 Flaws may not correspond to failures
      Margins  for safety?
      Latent flaws?
   New engineering practice                          Center for Devices and
                                                     Radiological Health

 Predictive Modeling
       Derive the risks (safety, costs, manufacturing methods,
        materials, etc)
 Usage simulation
      Look and feel, coexistence, training, maintenance

 Model based development
      Avoids building in defects, formally derives the system
       properties
 Model as a surrogate
      May remove clutter, and allow visibility, may provide a
       baseline.
  Examples of bio-informatic
  modeling currently in use                   Center for Devices and
                                              Radiological Health


 Safe leakage current limits
   Previously very little science involved
   Modeled on a whole body simulation


 Same with
   Safe temperature (contact, infusion, etc)
   Radiation transport in whole body
   Materials degradation
   Stent flexing
   Elution of substances from implants and food
    containers
  What do they have in
  common?                                  Center for Devices and
                                           Radiological Health


   It may not be safe, practical or ethical, for
    regulators to always use clinical methods to
    determine the acknowledged state of the art.




Maybe these are new clinical methods!