Prevention, detection and amelioration of adverse events in protocol-based decision support

The prevention, detection and amelioration of adverse events feature prominently in oncology protocols, the detailed plans for conducting clinical trials of therapies for the treatment of cancer. This paper summarises a safety review of such protocols prior to implementing OaSiS, a decision support system (DSS) in oncology [11]. OaSiS, shortly to undergo preliminary field evaluation in an oncology clinic, has been strongly influenced by the ONCOCIN [19], OPAL [17] and EON [16] family of computer systems developed at Stanford University in the last fourteen years. Like EON, OaSiS is a shell to be used with any oncology protocol provided it can be suitably represented. It does not support or critique the design of new protocols as do OPAL and DaT [21]. Even so, the safety review is still relevant to design tools like OPAL and DaT. OaSiS is being developed at the Imperial Cancer Research Fund (ICRF) in RED, a multi-partnered project funded under the UK Safety Critical Systems Programme. A safety critical system is one where operator error or system malfunction gives rise to serious injury, death, environmental damage or financial disaster, with possible legal liability. The safety related interests of RED partners include the assessment of safety critical software (Lloyd’s Register of Shipping) and the legal liability arising from safety critical systems (Masons Solicitors). Logical reasoning about safety issues is an important focus (ICRF and Queen Mary and Westfield College, University of London), as is the production of generic decision support system generator for safety critical systems (Integral Solutions Ltd). A more detailed account of the safety review can be found in [12] and [10] contains a more general discussion of safety and soundness in relation to expert systems.