I am a cancer survivor – survived a few times; sometimes it was cancer, sometimes it was not and sometimes they were just confused. What am I doing at an event on Demystification of AI. We have to start from the beginning to understand why I think AI can be a life line for people like me.
I first had a lump in my right breast when I was in college. It was the early 70s and I lived in Delhi. It was agreed that it was not malignant and could possibly be treated with homeopathy. Well, it just contracted and expanded but did not go away. A few months later the family surgeon decided to excise it – the view on the malignancy had not changed but the prevalent view was that these cysts could turn malignant. A small surgery and scar later, I carried on with life. But never is life the same again – they had severed some of my nerves.
Fast forward twenty years. I am the mother of two beautiful young boys 7 and 12 with a successful career. One sunny day in Bombay with the sea breeze gently wafting inland I walk into my GPs clinic to discuss my headaches and possibility of prescription glasses. I am travelling to London to read a paper at a conference. He introduces me to the new mammogram machine freshly installed and wants me to inaugurate it by being the first person to have a mammogram. I did it think it was a bit morbid at that time.
Anyway, I step on and I remember him asking me “Do you have pain in your right breast, Kamini”. NO not at all, I answer with fear creeping into my voice. I step back on one more time. He shows me the mammogram where he points to a fine network of lines and then to a denser cluster. He explains that the calcification (an indication of cancer) is just under a scar on my right breast that is not too recent. Next day we are at Tata Memorial Centre – the only and biggest cancer hospital then.
“Is it malignant”, I ask Dr Praful Desai the surgeon. He does not know, and he will only find out on the operating table or a little later from a frozen biopsy. But his long years of practice sure told him what it was. All the telltale signs were there. He will try and save my breast by doing a lumpectomy – according to the then protocol- but if the malignancy has spread and worse still it has metastasised and gone to other organs then he has to take a decision on the Operating table where I will be under anesthesia and therefore I must sign a form giving him authority to decide the course of my life.
It was malignant. The orthodoxy of the follow up protocol would put a religious zealot to shame. I am bombarded with radiation and chemotherapy for a year. They did well, as I am still here.
The next twenty years was life as usual or as much as it can be with all the drugs and radiation now ravaging my system. Of course, my life was frequently punctuated with annual mammograms, ultrasound scans, occasional CTscans and MRIs. But mostly fear. But they do assure me that if I survive the first 5 years, I am as good as new – with the same risk as the next person. I cross the first five years, then ten, fifteen and then
In the twentieth year, so forty years after my cyst, I wake up with a big lump in the same breast and almost at the same place. Twenty years of annual mammograms, scans etc. How did they miss this? My oncologist explains that this is a slow growing tumour and the machines do no pick up anything less than 7mm. I could have had it for 10 years or more. Now comes the most interesting bit of the story. Several tests later, the pathologist declares it is Stage 2 cancer. The earlier lumpectomy too was Stage 2. Two days later, I am told the pathology report has downgraded it to stage 1 and two days after that the pathologist says it is not cancer but just atypical cells and finally that it is DCIs that is Stage 0. Imagine my confusion, fear and anxiety? My oncologist and surgeon are now unanimous in pronouncing the protocol which is mastectomy. I learn that DCIs need not be treated but because of my history I must follow the path of mastectomy – complete removal of breast and possibly both if I have BRACA genes. I lose faith in the system and fly to Sloane Kettering in NYC where the pathologist says he has no doubt it is cancer. I spend two hours ascertaining why two pathologists of such repute would come to diametrically opposite views on the same sample. He explains that cancer diagnostics is not an exact science – based on appearance, number and position of the cells a pathologist would use his judgement to decide!!! And that one man or woman will decide my fate. Was it a single DCI or a multinodal cancer or DCI, I ask – they have no way of knowing – they will find out everything on the operating table.
This story ended well, I had a mastectomy and luckily it was DCIs and I needed no follow up chemo.
What I have told you is over forty years of my life. The subtext is the pain, the discomfort, sleepless nights, coping with life and depression, possibility of leaving without seeing my sons grow up and the fear, always the fear.
A year ago, my husband and I decided to support cancer – research, diagnosis and treatment and we looked for projects that would expedite research, with greater consistency and accuracy, minimum side effects and expense.
I read about a company sending drones strapped with medication into the body targeting cancer cells alone. This was no more science fiction. I knew then Technology was the answer.
I learnt that Healthcare is really data science. Of course, I had known it all along – One always hears that the best doctors are those that see more patients – diversity and scale helps predictive diagnostics. So, STEP IN AI. Analysis of large scale data by machine learning algorithms helps understand basic biology of cancer, early detection, treatment optimisation and help new drug discoveries.
I think, if they had information on my fibrosis and not excised it maybe it would not have set off a chain of events. If they had more evidence of the kind of malignancy, my lumpectomy perhaps would not have been followed by such drastic chemo and radiotherapy – all these have consequences but that is another story. And if they could predict that my DCIs would never turn malignant and die with me, I would not have needed the mastectomy. So many questions.
I am hoping that AI holds all the answers for the next gen
A lot is happening in this space –
CRUK, is funding research projects using AI across the entire spectrum –
For Early detection They are using data on consumer behaviour, lifestyle and health records
For Treatment optimisation Certain types of cancer have no standard protocol, so AI can help avoiding unnecessary surgeries.
For Drug discovery Bringing multidisciplinary data and machine learning. This brings down cost and time significantly.
Now I will try and bring this to life by a real case study. Peter is founder of Kheiron Medical, they are deploying AI for early breast cancer detection by doing what highly trained radiologists do but AI brings greater consistency and accuracy, it is faster and cheaper. In mammograms for breast cancer in the UK radiologists look at about 200 images in a day, 2min per image- 15mn pixels per patient. In under 2 min they must decide whether to tick the box or not. It is a very hard job, propensity to make errors significant and the cost very high. And radiologists have long years of training. So, no surprises here – no one wants to be a radiologist. This is not a problem for the UK alone.
Ideal number of radiologists is 10 per a 1000 population. UK is pretty close to 1, Japan is 2, Asia is 0-1. This is a worldwide problem.
Kheiron is also using AI for Cancer Treatment Tracking. Measuring response to a drugcan make the difference between life and death. An ineffective drug must be changed sooner than later.
In the not too distant future every second person will be touched by cancer. AI and these pioneers are the harbingers of hope for us.
I am certain that AI will play a huge role in winning this global war.