Objectif: élaborer un index prédictif de la prééclampsie basé sur les données les plus significatives de lanamnèse, de la clinique, de la biologie et des tests fonctionnels afin didentifier les patientes qui présentent un risque élevé de développer une prééclampsie (PE). Méthodes: Etude cas-contrôle comparant, en dehors de toute grossesse, des patientes ayant présenté, lors dune grossesse antérieure, une prééclampsie (n=101) avec des patientes dont la grossesse a eut un décours strictement normal, à parier pour la gestité, la parité, lâge, lassuétude au tabac et le délai depuis cette grossesse (n=50). Les paramètres analysés sont lanamnèse, lexamen clinique, la biologie sanguine (coagulopathie, trombophilie, hyperhomocystéinémie, vitamines B, marqueurs rénaux et vasculaires); et approches morphologie et fonctionnelle (système cardiovasculaire et rénal). Etude statistique de régressions logistiques a été appliquée à lensemble des données afin de nen dégager que les plus représentatives dans chacune des directions suivies par létude (anamnèse, biologie, épreuve fonctionnelle). Résultats: Lidentification des patientes à haut risque de prééclampsie peut être faite efficacement (88 % de sensibilité et 88 % de spécificité en se basant sur les trois index de risques construits respectivement sur les données anamnestiques et cliniques, sur les données de la biologie et sur les épreuves fonctionnelles. Conclusion: Il ny a actuellement aucun dépistage possible des patientes à risque de prééclampsie en dehors de la grossesse. Cette étude permet lélaboration dun index prédictif de la prééclampsie en préconceptionnel et par la même ouvre les portes à une prévention primaire.

No description available.

The thesis offers results of a research which has been conducted to look for and describe the typical personal characteristics of a change manager using Predictive Index tool and Kotter´s 8 steps process of change management. The thesis also offers set of practical recommendations for change management leaders as well as for the organizations.

Text in English / Although research paradigms and philosophies are highly regarded as frameworks and guides for choices of methods, application thereof is not evident. One of the reasons for the relatively limited application is the complexity and understanding surrounding paradigms and philosophies, making it hard for scholars to determine their stances and implications. This study describes a model for automatically predicting peoples’ paradigm and philosophical stance, including meaning, and their impact on research by helping the user determine the paradigm and philosophical stance closest to their beliefs. Paradigm and philosophical attributes are automatically derived from a set of structured questions that use information matching techniques. The development of a model for Research Paradigm and Philosophy Index (RPPI) follows a two-phase approach. The first phase involves automatic extraction of key indicators from a composed database that utilises an indexing scheme with different philosophies and associated implications. The second phase applies a matchmaking technique that automatically draws information reflecting the user’s attribute. This type of technology exists, but mainly in the dating and career matching fields. None exists for research paradigm and philosophical stances. The prototype system was designed and implemented to serve as a proof of concept, and was developed in Angular, using the Visual Studio Code, and Java, using Eclipse. The database was created using MySQL. The prototype system was designed and implemented to serve as a proof of concept due to the Intellectual Property nature of the product. Usability testing was conducted and results show that the participants agreed the system was simple, straight-forward to use, quite user-friendly and easy to learn, with easy navigation through menu items. / Computer Science / M.Sc. (Computing)

This paper investigates the investor trading behavior and the relationship between investor sentiment and stock returns. First we explore whether individual investors behave as the Disposition Effect stated — hold their losers too long while realize their winners too soon. Second, we apply four sentiment indicators — number of recommended stocks, margin purchase value, net fund redemption and odd-lot trade value — to examine relationship between investor sentiment and stock returns. We would like to see if past returns have anything to do with current sentiment, and if sentiment provides predictive power to future returns. First of all, from our analysis to over eight hundreds cash accounts trading records in two research periods, January to March and September to December in 2000, we found the Disposition Effect holds in average but not statistically. Second, the number of recommended stocks, weighted number of recommended stocks, margin purchase value, change in margin purchase value, net fund redemption and odd-lot trade value as proxies of investor sentiment are good at measuring the effect of past 4-week and 26-week returns on sentiment. Third, the margin purchase value, net fund redemption and odd-lot trade value provide predictive power to future 26-week returns in our study, which also implies there is likely underlying mean-reversion within half year during the research period. 　Finally, exploiting the change in margin purchase value as proxy of investor sentiment, we found the past 4-week returns volatility is inversely related with the indicator. That is, investors are scared on facing with high returns volatility.