Remarkable Omdena Journey: From a Newbie to Active Machine Learning Engineer
You just require that one opportunity to show the potential that you hold within; Omdena provided me this opportunity.
How did I come to know about Omdena?
“Self-belief and hard work will always earn you success.”
I recently completed my Bachelors in Computer Engineering in May 2021. During my undergrads I was always interested in developing algorithms for machines and having enough knowledge and understanding of Python Programming Language, I was attracted towards the solutions that Artificial Intelligence and Machine Learning could provide to society. I found out how useful and resourceful these solutions provided out to be and have these solutions can help humankind to make the Earth a better living place for living beings. Whether it be an Image processing problem, face recognition, classification, or even providing financial services, AI is for ALL. I was so much fantasized by the domain scope of AI and ML solutions, that I decided to start my very basic step to start acquiring the basic knowledge required. I joined several different renowned online courses offered by IBM, Stanford University, and the University of Michigan on the Coursera platform. After gaining enough theoretical knowledge, now was the chance to put my developed skills into action and test it out. I was in search of some online platforms that host AI and ML challenges that solve real-life problems and was rambling through various sites through the internet. Then I came across Omdena’s LinkedIn page. It just clicked me instantly that this would be the place where I can expose my skills as well as upskill. I felt this was one of the opportunities I should not miss out on. On March 8, 2021, I got selected for Junior Machine Learning Engineer for Omdena Flaskdata Project: Increasing Drug Safety by Detecting Anomalies in Clinical Data using Machine Learning. I was super excited to work with 50 collaborators joining from across the globe and working together as a team. This was the happiest day of my life and a big opportunity for me.
Problem Statement — Increasing the Drug Safety by Detecting Anomalies in Clinical Data using AI
In this challenge, the team had to build a service that would help to detect the anomalies found in the clinical data. Certain what’s and how’s to be addressed were: Can we rely on this data to make medical decisions? Can we quantify the risk? Can we rely on tech companies to use our clinical data for decisions that affect our lives without independent validation?
Keeping these questions in mind, the team gave the below SOLUTION:
A RESTFUL API was created which will help to automatically detect the anomalies found in the clinical data which will ease the accessibility of drugs and devices to the physicians. The team built an API to detect anomalies in structured clinical data (not free text or images) for two use cases: clinical trials and connected devices.
Use case 1: Clinical Trials — We used an ensemble approach to combine the detection of outliers in multidimensional data points, time-series drifts, and spikes as well as therapeutic-user-specified rules. For example, in psychiatric studies, some patients may report suicidal tendencies using the QIDS form using a mobile app. This is an example of a therapeutic-user-specified rule.
Use case 2: Connected Devices — Time-series data from connected wearable devices, watches, connected medical devices. Connected wearables may be standalone devices used by consumers or devices used in clinical trials to monitor efficacy, compliance, and patient safety. Use AD to monitor the clinical trial participants for adverse events/safety issues (for example consistently rising or dropping blood pressure may be indicators of a developing serious adverse event to the patient.
Things to keep in mind to successfully collaborate in the project
1. Speak up for yourself and take up the opportunity!
Omdena follows a bottom-up approach, which is not familiar to many Indian students. We are always bound to follow steps that our leading authorities ask us to follow. But Omdena following the bottom-up approach, there is no one there to assign you a task, you are free to take up any task that you feel follows within your knowledge and capacity. You must be dedicated enough to give your fullest efforts to give this task a meaningful and successful outcome. You have all the rights to be a task leader even though you have very little knowledge for it but have a mindset open to learn and extend your learning boundaries. There will always be people who would help you with all the support that you need to make this a successful task, it’s just you who needs to be open-minded and come out from the comfort zone to think for a solution out of the box. In the beginning, you might feel a little havoc of how’s and what’s but eventually, all will fall into place for the better. Also one of the major advantages of the bottom-up approach is more active collaboration and a sense of freedom of speech and to provide more innovative ideas.
2. Never be scared to ask for help and support
Omdena community has a very friendly environment to work in. People are always ready to help you with any of the roadblocks you face. You just need to be vigilant to ask your questions/doubts and anybody would be happy to help you. Also, Omdena’s slack has a #support channel, wherein you can take help from the members who have already participated in previous Omdena projects. At the same time, if you pertain the required knowledge to solve someone’s issue, never hesitate to offer help.
3. Be Pro-Active
Omdena community keeps track of the dedicated time frame to promised to provide them. You always need to actively participate in your preferred tasks across the project. Even if you possess very little knowledge of that task, no need to worry, you can still participate in that task, you can pitch in your ideas — they will definitely be listened off, always attend the meetings that are held on a daily basis and at the end, you would end up having a pool of knowledge.
4. Get yourself a little familiar with using everyday tools
Throughout the project timeline, you would come across many AI-ML tools that you would have to use. Apart from those tools, there are several tools such as Slack, Zoom meeting, Google Drive, Github, which you have some idea how to use them would make your journey more facile.
5. Acceptance and Acknowledgement
Believe in yourself and be allows self-motivated. If you get lost somehow in the middle of the journey — Try to acknowledge the opportunity that you have got, put all your efforts in, and don’t let all the hard work you have done so far go in vain.
I am so thankful to all the collaborators who actively helping and providing all the wisdom and right path to all the new fellow members to achieve success in the project.
Take-away from the Omdena Flaskdata Project:
There can be multiple solutions for one problem but not all solutions are optimistic and acceptable. Clinical data needs to be handled with utmost consciousness. Every record, every cell in the dataset has a meaning attached to a patient’s medical history. Sometimes you might not end up developing a generalized solution for all the types of clinical datasets. Every dataset needs to be understood properly. Performing an Exploratory Data Analysis would help a lot to understand the relations among the present data. It is always better to visualization the final outcome of any task, which would help in smooth transition among the various tasks.
Benefits from this Journey
· Being a part of this community taught me how to connect with people, share and cherish the ideas each and every person holds.
· My knowledge to address real-life problems has broadened. There are many real-life problems that can be solved using our technical AI skills.
· I understood the seriousness of understanding and knowing the data we are working with.
· I learned how to take the first initiative and never be afraid to express the thoughts I have.
· I learned different libraries used in an AI/ML project for different purposes such as for visualization of data we can use Plotly, Matplotlib, Seaborn, GGplot, and many more.
· Many new Machine Learning algorithms which I was not aware of earlier were used in this project such as Snorkel, Isolation Forest, Decision Tree. I get to learn how these algorithms work.
· Through this project participation, I got the experience of a working professional in the field of Data Science. I believe this experience would help me achieve better goals in my professional career.
An Opportunity worth not missing out on!
There will be people with different mindsets and with an amalgamation of different ideas, but the beauty of such a community is, eventually all agree upon one perfect idealistic solution which will be the most suitable for the project. I believe this is the new normal that I have learned after collaborating with the Omdena Community. Just one thing I believe whoever is beginning should keep in mind is: “Have Patience, eventually, everything will fall into its place.”
I am so grateful to the Omdena Community for providing me this opportunity to be a part of a life-changing project and would always look forward to many more active collaborations.
