Lamisa
Mahmud
Shady Business: Automatically Dimming Windows with an Adjustable Response Daniele Lucini, Raphael Kay, Joanna Aizenberg Emmanuel College, University of Cambridge | Chemistry | 2026 Indoor climate control is very energetically expensive, accounting for approximately 30% of global energy usage. Buildings with static elements cannot adapt to changing environmental conditions, thus rendering them inefficient and expensive as they are reliant on other devices, such as lights, to maintain a comfortable internal environment. Dynamic systems can broadly be classified as passive or active, depending on whether they respond to an external stimulus or user input. The former, whilst very energy efficient, cannot be decoupled from the external stimuli, and hence are not customisable. Conversely, active systems, although very adjustable to the needs of the user, are often expensive to run. We seek to create a dynamic system with a baseline passive response, that can be further adjusted actively at a low cost. Photoswitches are molecules capable of being in two states, which absorb different wavelengths of light. We incorporated photoswitches with a transparent, UV absorbing state, and a coloured, visible absorbing state into a film to account for the passive response of the material. When illuminated with broadband light from a solar simulator, an equilibrium between these two states is established, resulting in a loss of transmitted light. The percentage transmittance of visible light through the sample during irradiation was measured to quantify this passive response. We hypothesise that introducing fluidic filters containing either a UV or visible light absorbing dye will have a large effect on the position of equilibrium between the two photoswitch states and hence result in a significant change in transmissivity, even at low concentrations of dye. This would demonstrate that these films have an automatic response to light intensity, which can be altered by reversibly injecting fluidic light filters into a cavity. These systems would simultaneously have low energy costs, and a highly customisable indoor climate. Identifying Preventative Neurological Biomarkers Underlying Resilience to Childhood Maltreatment Lamisa Mahmud, Kyoko Ohashi, Martin Teicher
Abstract profile. Full document pending author claim.
Authors:
Lamisa Mahmud
Date Created:
Not specified
Course Title:
Professor:
Not specified
About Paper:
Childhood maltreatment (CM) is associated with heightened susceptibility to psychopathology. However, certain maltreated individuals exhibit no symptoms and consistently display a positive adaptation response to trauma, referred to as "resilience." Prior research has evaluated structural connectivity in the brain and findings revealed that specific preventative biomarkers allow them to compensate for the effects of CM. Yet, it's unknown how functional connectivity, which can reveal communication between brain regions, differs between the resilient and susceptible groups. This research aims to identify patterns of intra- and internetwork resting-state functional connectivity that may underlie resilience. Such patterns can then be induced in the susceptible brain to prevent or mitigate psychopathology. This study hypothesizes that resilient individuals display greater connectivity from the salience network (SN) to the default mode network (DMN) and reduced connectivity from the DMN to the SN. To examine this, 342 participants were placed into three groups- susceptible, resilient, or unexposed- based on their MACE (Maltreatment and Abuse Chronology of Exposure) scores and presence of clinical symptoms. Currently, resting state parameters are being analyzed using the CONN Toolbox to compare functional connectivity between all groups. Graph theory will then be used to calculate global network measures (e.g., global efficiency, small- worldness), followed by an analysis of variance and Tukey post- hoc tests to determine which groups in the functional network significantly differ from each other. Next, AI machine learning software will be utilized to identify a set of important nodes in the DMN and SN. Lastly, lasso regression and random forest regression will be conducted to predict group membership based on connectivity pattern. Combining these results with existing structural models could provide a comprehensive understanding of how the resilient brain differs from the susceptible brain and contribute to the development of more targeted interventions. Harvard Summer Undergraduate Research Village
Source:
Harvard / Neuroscience / 2026
Topics:
No topics listed
Co-authors:
Lamisa Mahmud