Why more data isn’t necessarily better – a cynical view on modern Health Tech

After reading a few reddit posts on med / health tech including people’s views on wearables (link), software compliance (link), vibe coding giving life altering lawsuits (link), programming (link). I have decided this is a necessary article to write. Though there are a lot of unanswered questions, they are very important and will certainly be revisited in future.

The world has the ability to generate more data than ever before. Especially with the advent of AI, there are more tools to analyse this data. There are increasing numbers of entrepreneurs creating startups in areas like toilet monitors (link, link ), smart watches or smart dog collars that capture GPS, temperature and monitor an individual or animal.

Though more data is being created, does this benefit us or our clinicians in making decisions? Or is it simply more unnecessary noise and low impact data?

For instance, the iKnife is very clear. During a typical operation involving tumour removal it takes 20 – 30 minute to for the histopathologist to make a decision on whether is this a tumour, boundary or normal tissue. Does the surgeon remove that tissue or is it safe to leave? The iKnife allows for real time decision making, with greater confidence providing the models have been verified allowing for reduced operating time.

Or in marathon running or training, smart watch health data can give answers to questions what is my heart rate? What cardiovascular zone am I in? Should I run slower, keep pace or run faster? How does my heart rate / pace compare to my last runs, am I improving?

These are clear actionable measurements that will ensure that individual uses that product for a long time.

There are levels to data.

Level 1: Foundation data. This is raw, uninterpreted facts and provides awareness but not context.

Level 2: Insight. Data given is contextualised and answers what happened or why but doesn’t provide action.

Level 3: Actionable data. This information specifically drives decisions, or immediate solutions to problems.

Level 4: Strategic data. This data alters long term goals, drives policy and behaviour. It is not just a single decision.

The majority of data from startups that are raising money in medical technology like smart collars or watches appear to be level 1 and 2. But it is also context dependant i.e. a GPS tracker on pets doesn’t help clinicians but if the pet has gone missing it allows for the owner to go find them.

I have thought and acted on Mass Spectrometry as being a diagnosis and prognosis tool assuming that from the data it provides clarity for action and strategy. But is that truly so. What exact problem am I solving? What data is necessary? What decision can be made? what behaviour will this alter?

The overall aim is for portable mass spectrometers to be diagnostic and prognostic tools. A diagnostic biomarker is a measurable biological indicator like a specific protein, or hormone used to detect or confirm the presence of a disease in a patient. It acts as objective clinical evidence to confirm a diagnosis and differentiate between conditions. A prognostic biomarker is one that indicates an increased (or decreased) likelihood of a future clinical event, disease recurrence or progression in an identified population. Examples of future events include death, disease progression, disease recurrence or development of a new medical condition.

Let’s take a care home, looking after elderly people who are unable to look after themselves due to a loss of mobility or mental deterioration. Having urinary checker for certain biomarkers like hydration or kidney function. Are they useful? For example, if they are being used in care home, an elderly patient with dementia may not know their hydration levels but an attentive carer would see that they haven’t had any water. If the device records dehydration, will they be provided immediately with water? This is a single action, but does it warrant the cost of the device. I couldn’t say so. Whereas a device that is able to track kidney function, track metabolism of kidney medication and for instance place the individual into a cohort like those who experience allergic effects or the drug is ineffective in. This is actionable data, that changes the behaviour to say this person should no longer take this drug or should swap to another one. Perhaps this saves money, as the patient no longer taking this drug or moves to a more effective or cheaper one.

Can portable mass spectrometers provide discrete levels for prognosis of a patient and so give actionable data that allows for actual meaningful behaviour change?

Let’s take longitudinal tracking. There are many questions here.

Does an elderly person want to know they have prostrate cancer, be reminded of it, track the progress of their worsening condition and their family see it as well? Would portable mass spectrometers even be something that carers want? What do carers want for their elderly patients? What do the families want? What do the patients want? What does the government want? What decisions are being made on this?

Same with at home use of a portable mass spectrometer, do people really want longitudinal tracking for their health? Can portable mass spectrometers really provide this type of monitoring? How exactly could and it be provided?

What are the hurdles for portable mass spectrometers?

Metabolic abnormalities caused by cancer can be detected in the blood, sputum, urine and other biological fluids via systemic or local circulation (link). However, there is a great challenge in biomarker reproducibility. A review that looked into non-nucleotide biomarker discovery protocols from body fluids in breast cancer diagnosis (link) found that the reported markers are often inconsistent and irreproducible across different studies and cohorts. There are many confounding effects including microbiomes, matrix, ion suppression and other effects. With analytical parameters sensitivity & specificity, calibration curve, accuracy and precision, stability, carryover, dilution integrity and reinjection reproducibility (link) are all things that must be validated.

For general cancer staging such as saying that this tumour is Stage II or Stage III, it would be able to. Due to the way cancer stages are anatomically and pathologically graded via tumour size, invasion, lymph node involvement, metastasis, histology, grade, molecular subtype and imaging findings (Fig. 1). 

Fig 1. Cancer metabolic markers obtained by liquid biopsy. Combining liquid biopsy with metabolomics can provide potential biomarkers for different aspects of clinical cancer management.

What is a reasonable aim for portable mass spectrometers.

The most reasonable aim would be narrow monitoring decisions. Here are two such examples:

Infected patient – What is the microbial ID or does it have antibiotic resistance associated profile present?

Newborn – What errors are in their metabolism pattern that is consistent with a treatable metabolic disorder?

What evidence is needed for portable MS to be a prognostic product?

Analytical validation: precision, accuracy, LLOD & LLOQ, carryover, interferences, matrix effects, stability, calibration, batch effects, user-to-user variability and ruggedness.

Clinical validation: A large prospective cohort showing that the MS-derived score products progression, recurrence, hospitalisation, toxicity or response better than standard care.

External validation: different sites, different operators, different instruments, different populations

Clinical utility: evidence that using the result changes the behaviour and improves outcome, or at least reduces cost, time, risk or unnecessary procedures.

Regulatory path: defined intended use, defined sample type, define patient population, quality system and comparator method.

Workflow proof: normal nurses, technicians, field users or patients can actually run the test without a PhD level mass spectrometrist.

When researching and looking at different aspects of medical technology like the iKnife. A question that always evaded me is why these technologies aren’t more prevalent. This list of evidence makes it abundantly clear why. It is a long tedious process with no clear proof that it will indeed be commercially viable without large investment and massively successful compared to producing mass spectrometers for research and pharmaceutical companies.