Compare these 3 drafts: (draft 1) The vibrancy of my mother’s life was evident from her Facebook page. As a tailor working for our family’s company, she would post sketches of her new designs or photos with her sisters. She would post videos of cooking with my sister or about the conversation we had at dinner. Most importantly, she would post pictures about my sister’s and my achievements. Things would keep running in that circle had I not noticed the changes in her language. “Celebrating Tuan’s ceretony at school.” It was late May when she made a spelling mistake and called my name wrong at the same time in a five-word sentence while posting the photos from my year-end ceremony to her profile. The joke kept me laughing, until we received the official diagnosis the next morning: Alzheimer’s disease stage 3.The shift was immediate; her confidence wore out even faster than her memory, and her Facebook page became lifeless. That’s when I spent countless hours creating a journal with her most regularly used words so she could reference it whenever she wanted to post. Every night, we would go over the journal together to work on her spelling. Over time, the notebook expanded to include names she used to know how to spell perfectly. My mother began to call me Anh again, the name she had bestowed upon me.The quest to find more brain exercises for her took me through many websites, until I stumbled upon the headline: “Alzheimer’s Among Top 10 Worldwide Diseases”. Determined to make an impact, I joined my school’s science research program, specifically the Brain Health Lab, where I explored using Natural Language Processing to detect Alzheimer’s. The first step was mastering the fundamentals of AI programming.Since cramming free tutorials on YouTube was overwhelming with floods of unseen terminologies, not letting my brain store any of them for more than a day, I sought distraction in my mom’s blog. Skimming through all the 42 posts updated in the past month showed me how the notebook helped my mother write again with joy and with fewer errors, I created a Quizlet module - my digital notebook - to keep record of the language of programming. When the word count in Quizlet shrank from 200 to 75% of accuracy in our predictive model’s first run, a sense of motivation filled me.For the subsequent weeks, repeated attempts to improve the model's accuracy hovered. Returning to my mother's profile, I hoped to see a new post, but was greeted with silence. I refrained from asking, until I noticed something other than Facebook on her phone screen - the Wordle interface. As my mother stopped monotonously memorizing the notebook and turned to solving word puzzles using Wordle, she had a chance to recall the words she learned in meaningful contexts. As her spelling mistakes became less frequent, I was inspired by this new learning style and applied “fine-tuning” to the model, adapting its settings to our specific task instead of using the trained parameters. The 85% accuracy of the new fine-tuned model printed on the screen, then to the manuscript we would represent at the school symposium.Throughout the last few months, my mother has had good days and bad days, but she has always persevered. Her perseverance has encouraged me to work towards what I want, even when it seems impossible. Joining the Brain Health Lab allowed me to use my passion for helping those, like my mother, with Alzheimer’s disease. With every step I’m taking in this path, the notebook keeps me standing still whenever I feel beaten up, and Wordle pushes me forward whenever I feel satisfied standing still.#“You have 1 new notification(s) from Facebook.”My mother posts to show the world what she loves. As the feelings in each photo from the symposium yesterday will live on to the future conferences I’ll be attending as a Ph.D., I’ll be making the world post about me, and let them know about my biggest love.
(draft 2) The vibrancy of my mother’s life was evident from her Facebook page. As a tailor working for our family’s company, she would post sketches of her new designs or photos with her sisters. She would post videos of cooking with my sister or about the conversation we had at dinner. Most importantly, she would post pictures about my sister’s and my achievements. Over time, I noticed slight changes in her language. It started as spelling mistakes she made on her Facebook posts and asking how to spell “dinner” or “ceremony”. I became concerned when she miscalled my name, using “Tuan”, instead of Anh or Tuan Anh. I knew that something was wrong. In May 2023 we received the official diagnosis: Alzheimer’s disease stage 3. The shift was immediate; her Facebook page became lifeless. Looking for a way to help her regain her confidence, I spent countless hours creating a journal with her most regularly used words so she could reference it when she wanted to post. Every night, we would go over the journal together to work on her spelling. Over time, the notebook expanded to include names she used to know how to spell perfectly. My mother began to call me Anh again, the name she had bestowed upon me. In my quest to find more brain exercises for her, I stumbled upon the headline: “Alzheimer’s Among Top 10 Worldwide Diseases”. Determined to make an impact, I joined my school’s science research program, specifically the Brain Health Lab, where I explored using Natural Language Processing to detect Alzheimer’s. The first step was mastering the fundamentals of AI programming. Since cramming free tutorials on YouTube was overwhelming with floods of unseen terminologies, not letting my brain store any of them for more than a day, I sought distraction in my mom’s blog. Skimming through all the 42 posts updated in the past month showed me how the notebook helped my mother write again with joy and with fewer errors, I created a Quizlet module - my digital notebook - to keep record of the language of programming. When the word count in Quizlet shrank from 200 to 75% of accuracy in our predictive model’s first run, a sense of motivation filled me. For the subsequent weeks, repeated attempts to improve the model's accuracy hovered. Returning to my mother's profile, I hoped to see a new post, but was greeted with silence. I refrained from asking, until one day, I noticed something different on her phone screen - the Wordle interface. As my mother stopped monotonously memorizing the notebook and turned to solving word puzzles using Wordle, she had a chance to recall the words she learned in meaningful contexts. As her spelling mistakes became less frequent, I was inspired by this new learning style and applied “fine-tuning” to the model, adapting its settings to our specific task instead of using the trained parameters. Never would I forget the 85% accuracy printed on the screen, which hardly fluctuated even in following attempts. Throughout the last few months, my mother has had good days and bad days, but she has always persevered. Her perseverance has encouraged me to work towards what I want, even when it seems impossible. Joining the Brain Health Lab allowed me to use my passion for helping those, like my mother, with Alzheimer’s disease, and gave me a glimpse into the research process. We are currently writing a manuscript to inform others of our approach and the accuracy of our model to detect Alzheimer’s disease based on natural language in narrative speech. The perseverance I learned from my mother will be necessary as my goal is to achieve a Ph.D. so I can continue to conduct research using AI to detect Alzheimer’s disease. AI has the potential to detect Alzheimer’s disease quickly and in an objective way, allowing for early detection and less cognitive decline. I hope that my research will allow for early detection of Alzheimer’s disease to prevent cognitive decline and enhance the lives of Alzheimer’s patients so they can live vibrant lives. draft 3 The research coordinator’s last words set fire to our school’s science symposium. Stepping onto the stage, I clutch a folded note in my hand; my heart pounds in my chest, with anxiety and happiness as my eyes skim through the note and stop at the final line that I didn’t write: “Love you, Tuan”.  My shyness fades as I spot my mother in the third row, her eyes glowing with pride and love.She is the reason why I’m here. #“Diagnosis: Alzheimer’s Disease stage 3.”As she blinks, the lodged words in my throat unravel, unleashing a rapid-fire supercut of the past three months.I frowned as everything fell into place. The diagnosis at the psychiatric hospital confirmed the fear.“Honey, how do you spell “dinner?””My mom asked as she was posting the photo of our meal to her daily blog. Not knowing it was the first sign, I laughed amusingly upon her joke, but soon stopped when I heard her greet me, with my middle name rather than first name, as “Welcome home, Tuan” after school.  Returning home from that moment, I rushed to my desk, armed with a Sharpie and a determination to create a lifeline for my mom. I wrote down the entire alphabet, compiling it into a notebook with footnotes detailing the correct spelling of her frequently-used words. Each night, turning down the chat screen with my friends, I found myself helping her spell. Over time, the notebook expanded to include names she used to know how to spell perfectly. Yet, my mother always reverted to Anh, the name she had bestowed upon me.In the search for more brain exercises for her, I stumbled upon a headline: "Alzheimer's Among Top 10 Worldwide Diseases.” Determination ignited. Joining my school's science research program led me to the Brain Health Lab, where I explored using Natural Language Processing to detect Alzheimer's.My journey began with mastering the fundamentals of AI programming. Since cramming up free tutorials on Youtube was overwhelming with floods of unseen terminologies, not letting my brain store any of them for more than a day, I sought distraction in my mom’s blog. As skimming through all the 42 posts updated in the past month showed me how the notebook helped my mother write again with joy and with fewer errors, I created a Quizlet module - my digital notebook - to keep record of the language of programming. The short span of memory made me feel like an Alzheimer’s patient, so I looked at a real one to overcome it. When the word count in Quizlet shrank from 200 to 75% of accuracy in our predictive model’s first run, a sense of motivation filled me.Moving forward from Quizlet, Wordle pushed things further. As my mother stopped monotonously memorizing the notebook and turned to solving the word puzzle, she had a chance to recall the words she learned in meaningful contexts. As her spelling mistakes became less frequent, I was inspired by this new learning style and applied “fine-tuning” to the model, adapting its settings to our specific task instead of using the trained parameters. Never would I forget the 88% accuracy printed on the screen, which hardly fluctuated even in following attempts. Finalizing on this model, we splitted up to start composing the progress report“Report submitted. Good luck preparing for your presentation.”#The moment I finish the concluding statement, the whole venue bursts into applause and congratulations. Before leaving the stage, I glance through the auditorium in pride before meeting a glowing, misty pair of eyes on the third row. She’s writing my name wrong again, but it’s okay. Today, what she needs to remember is how her perseverance inspired me to step on this stage. Tomorrow, as I delve further into the journey of helping millions of people with Alzheimer’s, the rest of this room will remember my name for her.