Study On Prognosis And Risk Stratification For Critical Illnesses

As a life-threatening complication of infection for critical patients, sepsis is associated with high morbidity and mortality rates despite the current advances in critical care. This is partly due to the lack of effective early diagnosis and risk assessment system for sepsis, which can help clinicians make appropriate and timely prognosises.Therefore, in this thesis, under the guidance of clinical medicine, the sensitive parameters for mortality risk prediction and severity evaluation of sepsis were screened. By the use of engineering technology, a prognosis model and a risk stratification model were both developed to realize the effective assessment of sepsis. Accordingly, the following tasks were accomplished:1. Designed and maintained a website which was used to collect electronic medical records (EMRs) of patients with sepsis from16hospitals in China.2. Analyzed the clinical parameters to obtain the specific ones that were most related to the mortality and risk stages of sepsis.3. Applied Monte Carlo simulation to determine the most appropriate models from naTve Bayesian classifier, support vector machine, and radial basis function network, depending on the statistical characteristics of the sample set.4. Constructed a prognosis model to assess the28-day mortality risk in sepsis patients.5. Established a risk stratification model to assign different labels to each EMR according to the disease severity of sepsis.6. Compared the prognosis ability of the newly constructed models with currently used APACHE II scoring system and the SOFA scoring system.The developed models are expected to provide a useful tool for the mortality prediction and risk stratification of sepsis, and this study offers an objective and effective method for clinicians to make accurate treatment decisions for sepsis timely and accurately.