While it is expected that the use of the medical decision-support systems would improve the safety, quality and efficiency of health care, there is limited evidence of this. The extent to which administrative efficiency is improved depends on multiple factors and a range of issues and challenges need to be addressed before health care professionals will use the systems more extensively; so the currently available systems have not yet been very successful and certainly their use is still not widespread and not established in daily routine. A variety of reasons may be responsible for this:

1. Lack of Accuracy: Those current systems that intend to cover a broad diagnostic domain of medicine generally lack diagnostic accuracy. This is mainly due to the levels of detail (e.g. diagnostic categories at the level of ICD-10 ) and completeness in the knowledge base. In contrast, systems based on detailed modeling of knowledge, resulting in good performance, are restricted to a relatively narrow field.
2. Lack of Transparency: In the era of evidence based medicine, the advice of ‘a machine’, functioning as a black box is unacceptable: an advice must be accounted on the basis of research published in the peer reviewed literature. The majority of conventional protocols and consensus guidelines also often fail to refer explicitly to the literature. Therefore, (diagnostic) advices suggested by a computerized tool should come with the appropriate references from the literature.
3. Users Attitude: In a subset of (potential) users there may be a misunderstanding about what computers can and cannot do for them. Generally, decision-support systems need intelligent and responsible users, who are able to interpret the advice given and estimate its merit. This however is not exclusively a matter of users’ attitude. Producers of decision-support tools should take this issue into account as well, especially when designing the user interface and deciding which facilities are needed.
4. Lack of Integration of Information: Patient-oriented decision support needs data from several sources. A decision-support system will generate new information (e.g. a diagnostic advice) through inference, using patient-specific information. Integration of information, multiple usability of patient data, integration of databases and knowledge bases are common problems when using a heterogeneous Hospital Information System (HIS). In practice, the completeness of patient information and the accuracy and level of detail of diagnoses stored in the HIS is often very poor.
5. Lack of a Controlled Terminology: This is a problem that even might not be solved completely in the near future. Most standard classification systems are at a general level, thus lacking the required detail, or specialized and therefore too limited to meet the needs for a broad decision-support system. Furthermore, there is not always a standard classification available, for instance for specific terminology used in text books.
6. Careful Introduction: Introduction of a decision-support system should be done as careful and thorough as is done for drugs that are new on the market. Oddly enough this tradition of careful introduction (and marketing!) is common in the field of therapeutics, but not quite as established for support tools in general and for diagnostics in particular. After introduction, the decision-support system will need constant monitoring of users’ needs and maintenance to keep up with the latest results of medical research.
7. The Need of an Integrated Clinical Workstation: The appropriate infrastructure and workstations are not yet available in all hospitals. Physicians will need on-line support during the implementation of the various functionalities of a reliable clinical workstation, which integrates all the required information.
8. Physician Barriers: Physicians can play a vital role in the success of most medical decision-support systems; however they can also create many barriers due to: desire for autonomy, physician personality and emotional issues, lack of trust of administration, failure of systems to clearly demonstrate better quality and outcomes, failure of organizations to provide effective incentives to use systems, lack of physicians leadership and involvement with project management, and finally, cultural differences to healthcare management.

In conclusion, modern medicine is in need of computerized decision aids both to meet its own high standards and to keep pace with the stage of development in other domains such as manufacturing or the services industry. Although decision support appears to be exceptionally suitable for the medical domain, computer aided decision making in medicine is still in its infancy. The development, implementation, assessment and further improvement of decision-support systems in medicine still need a lot of research.