http://hdl.handle.net/1721.1/7890 http://hdl.handle.net/1721.1/42338 Septic shock : providing early warnings through multivariate logistic regression models Shavdia, Dewang (cont.) The EWS models were then tested in a forward, casual manner on a random cohort of 500 ICU patients to mimic the patients' stay in the unit. The model with the highest performance achieved a sensitivity of 0.85 and a positive predictive value (PPV) of 0.70. Of the 35 episodes of hypotension despite fluid resuscitation present in the random patient dataset, the model provided early warnings for 29 episodes with a mean early warning time of 582 ± 355 minutes. Early goal-directed therapy (EGDT) in severe sepsis and septic shock has shown to provide substantial benefits in patient outcomes. However, these preventive therapeutic interventions are contingent upon an early detection or suspicion of the underlying septic etiology. Detection of sepsis in the early stages can be difficult, as the initial pathogenesis can occur while the patient is still displaying normal vital signs. This study focuses on developing an early warning system (EWS) to provide clinicians with a forewarning of an impending hypotensive crisis-thus allowing for EGDT intervention. Research was completed in three main stages: (1) generating an annotated septic shock dataset, (2) constructing multivariate logistic regression EWS models using the annotated dataset, and (3) testing the EWS models in a forward, causal manner on a random cohort of patients to simulate performance in a real-life ICU setting. The annotated septic shock dataset was created using the Multi-parameter Intelligent Monitoring for Intensive Care II (MIMIC II) database. Automated pre-annotations were generated using search criteria designed to identify two patient types: (1) sepsis patients who do not progress to septic shock, and (2) sepsis patient who progress to septic shock. Currently, manual review by expert clinicians to verify the pre-annotations has not been completed. Six separate EWS models were constructed using the annotated septic shock dataset. The multivariate logistic regression EWS models were trained to differentiate between 107 high-risk sepsis patients of whom 39 experienced a hypotensive crisis and 68 who remained stable. The models were tested using 7-fold cross validation; the mean area under the receiver operating characteristic (ROC) curve for the best model was 0.940 ± 0.038. Includes bibliographical references (leaves 87-89). Thesis (M. Eng.)--Harvard-MIT Division of Health Sciences and Technology, 2007. http://hdl.handle.net/1721.1/42217 Value creation through modernizing Chinese medicine Sun, Lizhe My first hypothesis in this thesis is that there is significant value vested in traditional Chinese medicine that can be captured by converting them into ethical drugs through scientific analysis, screening and validation. Further, holistic treatment is a key difference between traditional Chinese medicine and western-type chemical drugs, which makes Chinese medicine a very valuable category of knowledge. Using mixed formula is a primary method of treatment in Chinese medicine. It is the application of distinctive medical philosophies of Chinese herbal medicines in practices, reflecting the uniqueness and advantages of Chinese medicine. For example, there are 96,592 mixed formula recorded by "Dictionary of Chinese Medicine Mixed Formula" published in 1997. My second hypothesis in this thesis is that value can be created and captured, under the globalization context, from mixed herbal formulas for the mainstream world market with the aid of fingerprint technologies. To enter western markets as officially approved drugs through critical pathways, both scientific and regulatory, Chinese herb drugs must demonstrate sound evidence for safety and efficacy. I address in this thesis one of the central concerns of the pharmaceutical companies and FDA, that is, how quality control and material consistency is assured and how toxicity and drug kinetics of Chinese herbal medicines, either in its raw form, its purified form, its composite extract form or its mixed formula form, may be measured with reasonable scientific certainty and what would be the likely trajectory of further research. (cont.) My thesis research involves the following aspects: firstly, I characterize, by and through historical review and analysis, the formation of unique Chinese holistic medical philosophy to apply herbal medicines, particularly mixed herbal formulas, to systematically modulate the human body to prevent illnesses, to combat health problems and to restore balanced health; secondly, I performed a comparative study on the regulatory systems between Chinese SFDA and US FDA to provide insights on the trend of harmonic convergence of laws and regulations and challenges going forward, including collection and extrapolation of relevant statistical data; thirdly, I researched emerging fingerprint technologies to address the central issues of standardization, quality control, material consistency, safety and efficacy measurements of Chinese herbal medicines; fourthly, I performed data collection on major Chinese sources of published literatures and patent applications/grants for public and private medicinal knowledge formation, which may be viewed as a surrogate indicator for embedded economic value in the system, to compare trend and gaps between China and developed countries; and lastly, I presented three case studies of development of an-diabetic drugs from herbal sources, to illustrate how value may be created and captured through using modern technologies to tap into the accumulative knowledge base in herbal medicine. The thesis concludes that there are significant values to be captured, by and through cross-border collaborations under the globalization context, from Chinese herbal medicine. Both ethical single molecular entity (singleton) herb-derived drugs and mixed formula herb-derived drugs may be created going forward. Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2007. Includes bibliographical references (leaves 110-114). http://hdl.handle.net/1721.1/42216 Factors influencing the time for FDA review of medical devices Singh, Inder Raj, S.M. Massachusetts Institute of Technology Companies must receive marketing authorization by the Food and Drug Administration (FDA) before they can begin commercial distribution of a new type of medical device in the United States. The premarket approval application (PMA) is the process by which this occurs. Companies submit a PMA after they have completed laboratory test, animal studies and human clinical trials to demonstrate the safety and effectiveness of the device for a specific condition, or therapeutic indication. Despite legislation in the early part of this decade to reduce the timeframe for FDA review of PMAs, these timeframes continue to vary dramatically and unpredictably from months to many years. The focus of this thesis is to examine factors which influence this timeframe. Hypotheses about factors that impact PMA review timeframes were developed by analyzing the review process and through interviews with industry representatives and FDA officials. The following factors were evaluated: year of submission to FDA, size of firm seeking approval, presence of prior approved PMAs by firm seeking approval, product category, first-of-a-kind device, number of amendments, expedited review status, advisory panel review, unanimous advisory panel vote, and confirmation of primary efficacy endpoints in pivotal clinical trials. The year of submission was considered a control variable. The other factors fall into one of three categories: applicant characteristics, device characteristics, and process characteristics. Analysis was limited to PMAs received by FDA from 2000 through 2005. Two levels of analysis were conducted. (cont.) First, the directional impact of each factor on PMA review time was evaluated. Second, regression analysis was used to develop predictive models for PMA review time, in days, and to test which factors have meaningful associations when controlling for other factors. Factors that have highly statistically significant associations with longer review timeframes include: a larger number of amendments, and designation as an orthopedic device. Designation as an orthopedic device has a particularly dramatic impact on PMA review time. Orthopedic devices have a mean PMA review time of 647 days, 240 days longer (66% more) than the average for all other categories combined. Even after controlling for process, device, and applicant factors, the impact of an orthopedic designation remains large, increasing the review time by 175 days (p<0.01). In a univariate regression model, each additional amendment is associated with 20.2 additional days (p<0.0001) of review time. After controlling for other factors, each additional amendment is associated with 17.5 additional days (p<0.0001) of review time. Although the number of amendments cannot be known - or predicted - in advance of PMA submission, its significance (R-squared of 0.25 in a univariate regression model) in predicting PMA review timeframes reinforces the notion that quality - primarily of the dossier, in terms of its organization, clarity and completeness, but also of the adequacy of the underlying data to substantiate safety and effectiveness - is critically important to the achieving a shorter PMA review time. Only one factor has an association that in the opposite direction to that hypothesized. PMAs with an expedited review status have mildly longer PMA review times, by 37 days, than those that were not expedited. (cont.) This result can be explained in part by the larger number of amendments on these PMAs (corr=0.32). When controlling for the number of amendments and other important factors, an expedited review designation has a significant impact on PMA review timeframes in the opposite, but hypothesized direction - it shortens PMA review times by 146 days (p<0.01). Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2007. Includes bibliographical references (leaf 98). http://hdl.handle.net/1721.1/42215 Best care practices in anesthesiology : development and evaluation of an electronic feedback system to improve physician compliance with evidence-based practices Sarin, Pankaj, M.D. University of Rochester Recently, hospitals, regulatory agencies, and insurers have renewed their focus on improving patient care and safety. Outcomes based measures are being utilized and hospitals are being asked to report on whether patients are being treated according to a standard of care or a best practice guideline. As peri-operative physicians, anesthesiologists are able to evaluate and, to a great degree, affect the pre-operative, intra-operative, and post-operative course of a patient. However, several barriers exist. Although best practice guidelines exist, current models to risk stratify patients need improvement. Individual anesthesiologists currently have no uniform way to measure patient outcomes, either in an institutional or provider specific manner, and many treat patients based on anecdotal experience rather than on evidence based medicine. We addressed these issues through development of an electronic feedback system. The demonstration system targeted the problem of postoperative nausea and vomiting (PONV) in the ambulatory surgery patient population. Because performance of existing PONV risk prediction models was poor and could not be used for educational purposes, we created a new PONV risk prediction model and compared it against existing models. The new, improved risk prediction model was incorporated into an electronic system that gathered patient outcomes data related to best care practice and then fed back the information to care providers. After implementation of the electronic feedback system, we evaluated its efficacy in improving compliance with best care practices. Thesis (S.M.)--Harvard-MIT Division of Health Sciences and Technology, 2007. Includes bibliographical references (leaves 55-57).