Early detection and intervention for infants at risk of autism spectrum disorders

Autism is a developmental disorder that is characterized by abnonnal social interactions and communications as well as repetitive and restricted activities and interests. There is evidence of a genetic component, as 5% of younger siblings are diagnosed if their older sibling has been diagnosed. Autism is generally not diagnosed until age 3 at the earliest, yet it has been shown that early intervention for children with autism can greatly increase their functioning. Because of this, it is important that symptoms of autism are identified as early as possible so that diagnosis can occur as soon as possible to allow these children the earliest intervention. This thesis was divided into two parts. The first looked at the psychometrics of two proposed measures, the Parent Observation Checklist (POC), administered monthly, and the Infant Behavior Summary Evaluation (mSE), administered bimonthly, to see if they can be used with the infant population to identify autistic symptoms in infants who are at high risk for autism or related problems because they have an older sibling with autism. Study 1 reported acceptable psychometric properties of both the POC and IBSE in terms of test-retest reliability, internal consistency, construct validity and predictive validity. These results provide preliminary evidence that parent report measures can help to detect early symptoms of ASD in infants. The POC was shown to differentiate infants who were diagnosed from a matched group that was not diagnosed by 3 years of age. The second part of this thesis involved a telephone interview of parents who reported developmental and/or behavior problems in their high-risk infants that may be early signs of Autism Spectrum Disorder (ASD). During the interview, a service questionnaire was administered to see what interventions (including strategies recommended by the researchers) their at risk infants and affected older siblings were receiving, how satisfied the parents were with them and how effective they felt the interventions were. 3 Study 2 also yielded promising results. Parents utilized a variety of services for at risk infants and children with ASD. The interventions included empirically validated early intervention (e.g., ABA) to non-empirically validated treatments (e.g., diet therapy). The large number of nonempirically validated treatments parents used was surprising, yet parents reported being involved and satisfied, and thought that the services were effective. Parents' perceptions of their stress levels went down slightly and feelings of competence rose when they accessed services for their infants. Overall, the results of this thesis provide new evidence that parent-report methods hold promise as early detection instruments for ASD in at-risk infants. More research is needed to further validate these instruments as well as to understand the variables related to the parents' choice of early intervention for their at risk and affected children.

Buried Treasure: The Economics of Leak Detection and Water Loss Prevention in Ontario

On average approximately 13% of the water that is withdrawn by Canadian municipal water suppliers is lost before it reaches final users. This is an important topic for several reasons: water losses cost money, losses force water agencies to draw more water from lakes and streams thereby putting more stress on aquatic ecosystems, leaks reduce system reliability, leaks may contribute to future pipe failures, and leaks may allow contaminants to enter water systems thereby reducing water quality and threatening the health of water users. Some benefits of leak detection fall outside water agencies’ accounting purview (e.g. reduced health risks to households connected to public water supply systems) and, as a result, may not be considered adequately in water agency decision-making. Because of the regulatory environment in which Canadian water agencies operate, some of these benefits-especially those external to the agency or those that may accrue to the agency in future time periods- may not be fully counted when agencies decide on leak detection efforts. Our analysis suggests potential reforms to promote increased efforts for leak detection: adoption of a Canada-wide goal of universal water metering; development of full-cost accounting and, pricing for water supplies; and co-operation amongst the provinces to promulgate standards for leak detection efforts and provide incentives to promote improved efficiency and rational investment decision-making.

Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and deterministic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel metaheuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS metaheuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.

Feature Selection and Classification Using Age Layered Population Structure Genetic Programming

The curse of dimensionality is a major problem in the fields of machine learning, data mining and knowledge discovery. Exhaustive search for the most optimal subset of relevant features from a high dimensional dataset is NP hard. Sub–optimal population based stochastic algorithms such as GP and GA are good choices for searching through large search spaces, and are usually more feasible than exhaustive and determinis- tic search algorithms. On the other hand, population based stochastic algorithms often suffer from premature convergence on mediocre sub–optimal solutions. The Age Layered Population Structure (ALPS) is a novel meta–heuristic for overcoming the problem of premature convergence in evolutionary algorithms, and for improving search in the fitness landscape. The ALPS paradigm uses an age–measure to control breeding and competition between individuals in the population. This thesis uses a modification of the ALPS GP strategy called Feature Selection ALPS (FSALPS) for feature subset selection and classification of varied supervised learning tasks. FSALPS uses a novel frequency count system to rank features in the GP population based on evolved feature frequencies. The ranked features are translated into probabilities, which are used to control evolutionary processes such as terminal–symbol selection for the construction of GP trees/sub-trees. The FSALPS meta–heuristic continuously refines the feature subset selection process whiles simultaneously evolving efficient classifiers through a non–converging evolutionary process that favors selection of features with high discrimination of class labels. We investigated and compared the performance of canonical GP, ALPS and FSALPS on high–dimensional benchmark classification datasets, including a hyperspectral image. Using Tukey’s HSD ANOVA test at a 95% confidence interval, ALPS and FSALPS dominated canonical GP in evolving smaller but efficient trees with less bloat expressions. FSALPS significantly outperformed canonical GP and ALPS and some reported feature selection strategies in related literature on dimensionality reduction.

The influence of investigator disturbance on aggressive behaviour and breeding success of ring-billed gulls (Larus delawarensis) /

Increased losses of eggs and chicks resulting from human intrusion (investigator or other) into seabird colonies has been well documented. In 1990/91, I studied the effects of investigator disturbance on aggressive behaviour and breeding success of individual pairs of ring-billed gulls nesting at two colonies near Port Colborne, Ontario. The insular colony was on an artificial breakwall, associated with the Welland Ship Canal, approximately 1 km off the north shore of Lake Erie. The mainland colony was adjacent to the canal approximately 1 km east of the breakwall. The frequencies of adult threat and assault behaviours, chick movement and adult attacks on chicks were recorded by continuous scan sampling 30 min prior to, 30 min during and 60 (2 X 30) min after investigator disturbance. The frequency of threat and assault behaviours increased during the period of investigator activity in the colony while the duration of wingpulls and beakpulls decreased. Significantly more chicks ran ("runners") from their natal territories during disturbances and "runners" were more frequently attacked than "territorial" chicks. No chicks were fatally attacked during disturbance and "runners" returned to their natal territories quickly after disturbance. Breeding success was determined for pairs nesting in study plots subjected to two levels of disturbance (normal and moderate). The disturbance level of each plot differed in visitation frequency and activities performed on each visit. Investigator disturbance had no effect on the hatching success or fledging success (taken as 21 days of age) of ring-billed gull study pairs at either colony.