Text Mining and Patient Severity Clusters - Text Mining Techniques for Healthcare Provider Quality Determination - page 320

Figure 24. Predictive model using memory based reasoning

a lower mortality rate should rank it higher compared to #7, if we can demonstrate that the patients in
these two hospitals have approximately the same levels of severity.

usIng PatIent severIty to PredIct outcomes for InterventIon

Unlike the patient severity indices defined in Chapters 5-7, we can use Text Miner to predict more specific
patient outcomes. We do not need to restrict attention to patient mortality. We demonstrate how such a
prediction can work by examining the occurrence of resistant infection.

Table 16. Text analysis for cardiovascular surgery to compare ten hospitals

Cluster Codes Translation Frequency

1 2766, 276, 2851,

285, 410, 4273

Fluid overload, Disorders of fluid, electrolyte, and acid-base balance, Acute posthemor-
rhagic anemia, Other and unspecified anemias, Acute myocardial infarction, Atrial fibril-
lation and flutter

323

2 272, v158, 4111,

401, 411, 2720

Disorders of lipoid metabolism, Other specified personal history presenting hazards to
health, Intermediate coronary syndrome, Essential hypertension, Other acute and subacute
forms of ischemic heart disease, Pure hypercholesterolemia

477

3 412, 413, 4130,

v 4 5 8 ,

Old myocardial infarction, Angina pectoris, Angina decubitus, Other postprocedural status,
Essential hypertension, Other and unspecified hyperlipidemia

112

4 0 1 ,

2724

4 997, 9971, 4273,

3051, 496, 5119

Complications affecting specified body systems, not elsewhere classified, Cardiac complica-
tions, Atrial fibrillation and flutter, Tobacco use disorder, Chronic airway obstruction, not
elsewhere classified, Unspecified pleural effusion

320

5 410, 318, 428,

4 2 8 0 ,

Acute myocardial infarction, Other specified mental retardation, Heart failure, Congestive
heart failure, unspecified, Subendocardial infarction, Acute posthemorrhagic anemia

234

4 1 0 7 ,

2851