Human Values in Computer Diagnosis
I would like to begin with two examples of how human values might be added into the calculus of computer diagnosis and clinical management. The first issue concerns the implications for outcome when a particular diagnosis is made. Not all diagnoses are made in the same way. Some have significantly adverse implications if they are present and others do not. Take, for example, a case of a middle-aged man with an episode of hematuria following mild abdominal trauma. There are a number of possible diagnoses, each with an associated outcome. The first option is a condition with no adverse outcome for the patient, for example, the presence of a solitary renal cyst. The primary reason for making such a diagnosis is that the cause of bleeding is identified and other conditions do not have to be considered. The patient is not directly benefited but is protected from future concern and the cost and risk of further examination. A second possibility is a condition with a bad or even fatal outcome for which nothing can be done, for example, an advanced pancreatic carcinoma which has now invaded the kidney. A third intermediate possibility is that a condition is found that is serious and treatable, or even curable. An example of this would be an early resectable renal cell carcinoma. From the viewpoint of the patient, diagnosis number three is the most important one to make since it provides an opportunity to save the individual from a potentially fatal disease and restore health.
A computer assisted diagnostic workup therefore should be directed towards making or excluding diagnoses such as number three — serious diseases that can be treated. One might argue that diagnosis number two deserves also to be made as a matter of priority because of the need to prepare the patient for palliative therapy and eventual death. But this situation should not receive the same priority as an urgent or emergent correctable condition. For a common condition such as hematuria, it should be possible to gain some consensus from the medical profession as to which conditions should be prioritized in the diagnostic scheme to provide maximum patient benefit. Indeed, some decision-making programs currently exist, for example, a pharyngitis protocol that focuses upon treatable conditions with potentially serious outcomes, i.e. streptococcal pharyngitis[l]. Many diagnoses in medicine, however, such as lupus erythematosis have highly variable clinical presentations and are not amendable to easy categorization as conditions with correctable outcomes. But by adding into the computer additional information about the lupus, for example the involvement of the brain versus the skin, the diagnostic and prognostic implications for the patient may be made more precise. Other conditions may have little import for the physician but great meaning for the patient. For example, a mild trichomonas infection might be considered by the medical profession as a non-serious cause of dysuria but it may have serious social repercussions for its victims since it is categorized as a sexually transmitted disease. Or, the diagnosis of early Alzheimer’s Disease creates few treatment options for the physician but has major, potentially devastating implications for the patient and their family. If the medical system is to be truly responsive to the needs of the patient, then a shift from values important primarily to the physician to those also of the patient is imperative. It may be possible to collect important values or concerns from collective groups of patients with certain diseases and weigh them for relative importance. If certain values such as associated pain and suffering, functional incapacity or even economic cost can be listed and quantified with satisfaction, it might be possible to add these into the traditional computer calculus. If some values prove too intangible and idiosyncratic, then the computer might simply encourage the physician to address and measure these issues with the individual patient.
KeywordsLaryngeal Carcinoma Streptococcal Pharyngitis Collective Group Variable Clinical Presentation Advanced Pancreatic Carcinoma
Unable to display preview. Download preview PDF.