Continued from my previous blog Loss function To make better predictions, we need to know how good or bad our model is performing. Loss function does exactly that. Based on the prediction, we compare it to the estimated value and check how much loss has happened in predicting right. This can be done using the function loss = nn.MSELoss(). MSE is mean squared error which makes the difference in the loss by squaring them Next, comes the optimizer. With a loss function, we can tell how well (or poorly) our neural network is performing. To improve on our model’s performance, we need to adjust the weights and biases (Read my linear regression blog understand weights better).

Building a Machine Learning model involves the following steps. Define what your model will do Collect data Choose model (Should read more about this) Train model on your dataset Evaluate model Tune and deploy model Tensors Pytorch is a python library which helps in building an ML model. FIrst, lets talk about tensors. Imagine a spreadsheet which contains rows and columns of data, tensors are more complex which can hold data in several dimensions. The common term used is “containers” that holds numbers in a structured way. Why is it important to convert data into tensors? it helps in easier computation of multi-dimentional data, it has a standardized format as they can represent image, numerical, text data in a consistent format, helpful in applying mathematical operations such as linear algebra.

I’m not a huge superhero fan neither am I a great artist. The animation in spiderverse just blew my mind. I’ve added 2 digital art work insipired by the movie and I have atleast 10 more sitting in my drafts, unfinished. Just wanted to document my effort.

Statement of Purpose As an embryologist I am in a unique position, giving hope to patients. My primary role involves performing procedures like gamete manipulation, embryo culture, embryo biopsy for preimplantation genetic screening/testing, counseling patients about their prognosis, and managing my department’s operations. It brings me so much joy seeing patients embrace babies after a successful IVF(In-vitro fertilization) cycle and when cancer patients secure their reproductive potential by freezing their eggs or ovarian tissue. My job is incredibly fulfilling. At the IVF lab, I make important decisions based only on the embryo’s morphology. However, this only offers very limited information on the embryo’s developmental capacity or ploidy status.

As a part of my part of my python programming course I have an assignment. I can pick a python library of my choice and give a demo. (Its a group project involving 3 people) We chose PyTorch as our library of interest. I have been reading since yesterday about PyTorch. Codecademy is a super great resource that I found online. I’ve been using it a little over a year now and its amazing. For someone who knows nothing about programming, its a great place to start. I took this really short course on codecademy on PyTorch and neural networks. Here’s what I learned.

A lot has happed in the past 2 months of my life. I married the love of my life. I moved to a new country and am a student again! Working on impactful research has been a longing pursuit since them past 4 years of my life. And I’ve only kept at it at every step of the way. Looking at where I am now and back at the past 4 years, it all makes sense. I am currently a master’s student at NYU studying Biomedical Informatics. NYU is one of the top medical schools and they are doing really good research.

Accessible Healthcare vs Affordable Healthcare Most individuals need to be more positive about visiting a doctor. The strong belief of a lot of people in my country (India) is that hospitals are there to cheat and once you visit them, they will be charged a hefty amount and asked to take multiple tests which might be unnecessary. Although I agree with this, I have a different opinion about healthcare services in India. As an Embryologist I have worked in healthcare for the past 5 years. Healthcare is certainly expensive. Since I am an Embryologist I can break down the expenses of an IVF cycle.

I work as an embryologist. I am the super scientist that makes IVF babies (test tube babies, logically petri dish babies but, nvm). IVF is more than an invasive procedure. There is so much emotion involved. Couples come to me (us) after multiple failures of achieving a pregnancy naturally in hopes of conceiving and having a baby (immediately in the first attempt). Most often these couples have a tragic history of infertility for several years and it usually takes more than one IVF attempt for these couples. You might wonder why people go through such an emotionally and physically draining procedure, one might just adopt.

Managing an IVF lab is no task at hand. With every waking day, I only pray that I do not come across any unsolvable problem. Right from the paperwork to handling gametes one needs to be highly vigilant and especially in the clean rooms such as the IVF lab. Handling stock solutions/culture media. There are 2 types of media used for IVF. The CO2 dependant and CO2-independent. All media are to be used when at 37.0 degrees. The CO2-dependent media requires equilibration. The general practice is to aliquot the media necessary for the next day’s case and keep it in the CO2 incubator for overnight equilibration.

Hey there! I know it has been a long time. Things have changed in 2 years. I am no more a student. I am an embryologist now, working in Manipal hospital Bangalore. It has been 6 months since I took up this job and it is excellent. I’m managing the lab end to end right from documentation, and QA QC to ICSI, and embryo transfer. I hope to write more here about my lab experiences and what I am reading.