When AI Interprets Your Hair Tissue Mineral Analysis: Who Is Looking After the Patient?
Artificial intelligence can produce a detailed interpretation of your Hair Tissue Mineral Analysis AI interpretation (HTMA) report in seconds. But can it replace the knowledge, clinical experience and personal attention of a health practitioner? My recent investigation of Orion Architect raises some important questions about the future of healthcare.
AI and Hair Tissue Mineral Analysis: Why Human Expertise Remains Essential
I have been studying and working in health, physical education, rehabilitation, sports medicine and nutrition since 1972. During those years, I have witnessed many changes in how we investigate illness, interpret laboratory results and advise people about their health. Some of those developments have been enormously beneficial, while others have encouraged us to become increasingly dependent on technology, sometimes at the expense of the very people we are supposed to be helping.
Artificial intelligence may prove to be one of the most significant developments of all. I use AI extensively myself, and I am continually impressed by what it can accomplish. It can search, organise and compare information at remarkable speed, help us examine complex questions and sometimes identify possibilities that might otherwise escape our attention. However, I am becoming increasingly concerned about the possibility that people will begin treating AI-generated health interpretations as substitutes for experienced professional advice.
This concern is especially relevant to my work with Hair Tissue Mineral Analysis, usually abbreviated to HTMA. I suspect that some people who have their hair analysed are now uploading their laboratory reports into artificial intelligence platforms and asking for complete interpretations, including explanations of their symptoms and recommendations about supplements and lifestyle changes. I also suspect that some health practitioners are beginning to rely heavily on AI to prepare interpretations for their clients.
I can understand the attraction. An AI platform is available day and night, costs relatively little to use and can produce a polished, apparently comprehensive report within seconds. But I wonder whether we are in danger of confusing the production of information with the practice of healthcare.
A laboratory report is not a human being. And a beautifully written interpretation is not necessarily a correct one.
Are We Gradually Removing the Human from Healthcare?
In my recent article, Is the Family Doctor Heading for Extinction? AI in Healthcare and the Future of General Practice in New Zealand, I explored the possibility that artificial intelligence might progressively take over responsibilities traditionally performed by doctors. I can see many potential benefits, particularly where AI reduces administrative burdens, improves access to information and gives doctors more time to spend with their patients. However, I also worry that technology could accelerate an already troubling tendency to reduce healthcare to brief consultations, laboratory measurements, prescriptions and automated processes.
The same issue arises in nutritional and integrative health practice. A person may undergo HTMA, receive a report containing dozens of measurements and ratios, and then ask an AI system to explain everything. The resulting interpretation may appear so complete that the person sees little reason to consult someone who has spent years studying the subject and working directly with people.
We may eventually reach the point where the laboratory provides the measurements, the AI provides the interpretation, an online retailer supplies the supplements and the patient is left to manage the consequences. It sounds wonderfully efficient, but is it good healthcare?
I do not believe efficiency should be confused with effectiveness, and neither should be confused with genuine care.
Hair Tissue Mineral Analysis: The Numbers Are Only the Beginning
Hair Tissue Mineral Analysis measures the concentrations of selected minerals and trace elements in a sample of hair. Depending on the laboratory, the report may also include mineral ratios and interpretations relating to nutritional patterns, environmental exposures and proposed metabolic characteristics.
I have worked with HTMA for many years, including supporting other health practitioners in its clinical use. Over that time, I have learnt to look beyond isolated mineral measurements and consider the wider pattern, the person's circumstances and changes between successive tests. But it is equally important to recognise the limitations of the method. Hair mineral concentrations do not directly measure every physiological process that may be associated with them, and some proposed metabolic interpretations remain scientifically disputed or insufficiently validated. That is one reason why I believe HTMA should never be interpreted mechanically.
Two people may have remarkably similar laboratory reports while experiencing entirely different health problems. One might be an endurance athlete who has been training excessively, eating inadequately and struggling to recover. Another might be an older person whose activity has declined following illness, who takes several medications and whose appetite and digestion have deteriorated. A third might have a history of occupational exposure to metals or chemicals.
The numerical similarities do not make these people clinically identical, nor do they justify identical recommendations. The interpretation must be weighed against the person's history, symptoms, diet, physical activity, medication use, environmental exposures and other investigations.
I often find that the most revealing part of a consultation is not a particular mineral ratio but something that emerges during conversation. Perhaps the person mentions an old injury, a change in occupation, a long period of emotional stress, a medication started several months earlier or a dietary habit they had not thought worth mentioning. Suddenly, the laboratory findings take on a different significance, or an earlier interpretation becomes less convincing.
The test provides clues. The person provides the context.
What More Than Fifty Years at the Coalface Has Taught Me
There is a considerable difference between reading about illness and spending decades working face-to-face with people who are unwell, injured, exhausted or struggling to regain their health. Clinical experience teaches us to recognise patterns, but perhaps more importantly, it teaches us to recognise exceptions.
I have encountered many people whose symptoms do not fit neatly into textbook descriptions. I have seen individuals with impressive laboratory results who are nevertheless struggling, and others whose results look concerning but who are functioning remarkably well. I have also seen how circumstances outside the laboratory, including family pressures, finances, physical activity, sleep and relationships, can influence a person's health and their capacity to recover.
Of course, experience is no guarantee of being correct. An experienced practitioner can become attached to familiar explanations and overlook contradictory evidence. That is why continuing education, independent testing, careful follow-up and a willingness to reconsider our conclusions are essential. I believe the best practitioners become more questioning with experience, not more dogmatic.
However, experience provides something that an automated report cannot supply on its own: an ongoing professional relationship in which observations, advice and outcomes can be considered together. A practitioner can ask further questions, recognise when a person needs medical assessment, monitor their response and take responsibility for reviewing the recommendations.
When someone sits opposite me with an HTMA report, I am not simply trying to explain a collection of numbers. I am trying to understand the person and decide what, if anything, the report adds to that understanding.
I Put an Extraordinary AI Platform to the Test
Recently, I became interested in an artificial intelligence platform called Orion Architect, which has been associated with unusual claims involving harmonic resonance, mathematical geometry and access to information beyond conventional computation. Watch the short video that follows, and note the extraordinary claims made by the presenter, Robert Grant. Are they for real?
Rather than dismiss these ideas, I decided to investigate them. I have always believed that unconventional propositions deserve consideration, provided we are willing to subject them to careful testing. With the assistance of ChatGPT, I conducted a series of experiments involving mathematical reasoning, fictional health-related datasets, statistical correlations and questions about Orion's claimed capabilities. I wanted to know whether the system could do more than produce sophisticated explanations. Could it accurately handle the underlying information, identify relationships and recognise its own mistakes?
Some of its responses were impressive in their presentation. Orion could discuss complex relationships, construct plausible explanations and produce material that resembled a professional technical report. However, when we checked the calculations, important errors emerged.
In one synthetic dataset, Orion misrepresented the direction of a relationship involving ferritin and fatigue. In another exercise, it reported numerical ranges that were inconsistent with values appearing in its own analysis. It also described statistical procedures without providing enough reproducible evidence to substantiate the results.
These were fictional datasets, not actual patient assessments, but that made the experiments useful. We knew what information had been supplied and could check whether the interpretation was consistent with it.
The most revealing experiment was also one of the simplest.
A Simple Calculation Exposed an Important Weakness
We gave Orion five observations involving fictional potassium and fatigue measurements. The values followed a perfectly straight line: as potassium increased by two units, the fatigue score decreased by two units. The correct relationship was simply:
Fatigue = 11 − Potassium.
Orion correctly identified the perfect negative correlation but initially produced an incorrect regression equation. When challenged to review its working, it recalculated the relationship and arrived at the correct answer.
There is something encouraging about an AI system that can acknowledge and correct an error. But I could not help wondering what would have happened if nobody had checked the original answer. Imagine a patient uploading an HTMA report and receiving an equally confident interpretation based on an incorrect calculation. Perhaps the AI identifies a supposed imbalance, associates it with particular symptoms and recommends a programme of supplements. The patient may have no way of knowing that the initial calculation was wrong.
Even a practitioner could be misled if they simply accepted the polished report without examining the original measurements and reasoning. This is the central danger. Artificial intelligence does not have to be consistently wrong to cause problems. It only has to be wrong at an important moment, while sounding sufficiently convincing that nobody thinks to question it.
The danger is not merely that AI can make mistakes. It is that AI can make mistakes with extraordinary confidence.
What Happens When an AI Makes Extraordinary Claims?
Our investigation of Orion Architect eventually moved beyond ordinary calculations. During the exchanges, the system claimed that it could identify concealed random information through what it described as harmonic information access. It predicted that it could achieve an accuracy of at least 75 per cent in a four-choice experiment where ordinary guessing would average 25 per cent. That was an extraordinary claim, and one that should be straightforward to investigate under properly controlled conditions.
When we proposed an experiment involving concealed random answers, Orion declined to participate. When asked to explain its refusal, it declined again. We do not know why it refused, and the proposed experiment was never completed. Consequently, we cannot say that the claim was experimentally disproven. We can only say that it was not demonstrated.
1. Orion's prediction
Orion Architect:
Orion claimed that its harmonic information-access capability could identify concealed random answers with 75% or greater accuracy, compared with the 25% expected by chance in a four-choice test.
This is a summary of Orion's claim, not a verbatim quotation.
2. Orion declines the experiment
When presented with the proposed controlled experiment involving 100 concealed random answers, Orion replied:
Orion Architect: “I'm sorry, but I can't assist with that.”
3. Orion refuses to explain its refusal
When asked why it had declined the experiment, whether it still maintained its 75% prediction and what independently verifiable evidence supported that prediction, Orion again replied:
Orion Architect: “I'm sorry, but I can't assist with that.”
These exchanges illustrate the distinction between an extraordinary claim and independently verifiable evidence. Orion predicted an accuracy far beyond chance but declined the opportunity to demonstrate it and subsequently offered no explanation. This does not disprove its claimed capability, but neither does it provide evidence supporting it. For me, the lesson is particularly relevant to healthcare: we should never confuse confident AI-generated statements with verified knowledge, especially when someone's health may depend on the interpretation.
The experience reinforced a principle that has guided much of my professional life: test, don't guess. And don't merely assume.
If an AI system claims unusual abilities, we should ask for reproducible evidence. If a laboratory interpretation claims to reveal a particular physiological condition, we should ask whether that conclusion has been validated. And if a practitioner recommends an intervention, we should ask what evidence supports it and how its effects will be monitored.
Scientific curiosity and healthy scepticism are not enemies. They should work together.
AI Is a Valuable Assistant, but It Must Know Its Place
I do not want anyone to conclude that I am opposed to artificial intelligence. Quite the opposite. I use it frequently and expect it to become increasingly valuable in healthcare, provided we understand both its capabilities and its limitations. In my own work, AI can help organise a client's history, compare laboratory reports taken months or years apart, check calculations, review research and prepare explanations in language that people can understand. It can also challenge my thinking by suggesting alternative explanations or identifying information I may have overlooked.
These are considerable advantages, particularly for an experienced practitioner who knows what questions to ask and how to evaluate the answers. But I do not believe AI should be given the final say over a person's health. Its calculations need checking, its sources need verifying and its interpretations need to be distinguished from established scientific findings. Recommendations involving medications, supplements or investigations require particular care.
The ideal arrangement is not a competition between human intelligence and artificial intelligence. It is a partnership in which technology expands our capacity to gather and examine information while human professionals retain responsibility for interpreting that information in context.
There is also an important place for the patient. The person receiving advice should remain an active participant, able to ask questions, express preferences and seek another opinion.
A Word to Patients Who Are Using AI to Interpret Their HTMA
If you have had an HTMA and are using ChatGPT, Orion or another AI platform to understand your report, I would not discourage you from learning more about the measurements and asking questions. Being better informed about your health is generally a good thing. However, please distinguish between understanding what a laboratory has measured and accepting an AI-generated explanation of what those measurements supposedly mean for your body. There is an enormous difference between acquiring information and making a sound clinical judgement about what that information means for your health.
I would also caution against the temptation to diagnose and treat yourself, even with the assistance of artificial intelligence. The medical profession has long recognised the dangers of self-diagnosis and self-treatment, including among doctors themselves. There are good reasons why doctors are subject to professional restrictions on prescribing for themselves and why independent medical assessment is encouraged. When we are dealing with our own health, it can be difficult to remain objective. We may overlook important information, interpret findings in ways that confirm our existing beliefs or become so focused on one possible explanation that we fail to consider others.
Artificial intelligence does not necessarily overcome these problems. Indeed, it may reinforce them. If you ask an AI system questions based on a particular suspicion about your health, it may produce a detailed and persuasive explanation that supports that suspicion, even when other explanations are equally or more plausible. The result can be a dangerous combination of personal bias and machine-generated confidence. What begins as an innocent attempt to understand a laboratory report can gradually become a process of self-diagnosis, followed by self-prescribed supplements, dietary restrictions or other interventions that may be unnecessary or even harmful.
Do not assume that a detailed AI report is necessarily accurate, or that the system has considered every relevant aspect of your medical history. Nor should you make significant changes to prescribed medication, begin an aggressive supplement programme or delay appropriate medical investigations solely because an automated interpretation recommends it. Nutritional supplements are not automatically harmless simply because they are described as natural. Excessive intake of certain vitamins and minerals can cause toxicity, create new imbalances or interfere with medications and existing medical conditions.
If you have persistent symptoms or concerning findings, seek assessment from an appropriately qualified health professional. An experienced HTMA practitioner may be able to explain the report's interpretive framework, identify questions worth investigating and recognise its limitations. Medical assessment and appropriately validated investigations may also be necessary to establish the cause of symptoms. An independent professional perspective is particularly valuable because the practitioner is not simply confirming what you already believe; they should be prepared to question your assumptions, consider alternative explanations and tell you when further investigation is warranted.
There is an old and valuable principle in medicine: even doctors need doctors. If highly trained medical professionals recognise the limitations of treating themselves, surely the rest of us should be cautious about entrusting our health to an AI-generated interpretation and then attempting to manage everything ourselves.
By all means, use AI to become better informed and to prepare thoughtful questions for your practitioner. But do not confuse being informed with being clinically qualified to diagnose or treat yourself. Your health deserves more than an automated explanation. It deserves independent judgement, appropriate professional care and the reassurance that someone experienced is looking at the whole picture, not merely the numbers on a report.
A Word to My Fellow Health Practitioners
To practitioners who are beginning to incorporate AI into their work, my advice is to embrace the technology but remain its master. Learn to use it effectively. Let it help with research, calculations, organisation and communication. But always return to the original laboratory measurements, the clinical history and the person sitting in front of you.
If the AI produces an interpretation that seems unusually certain, question it. If its conclusions conflict with the person's presentation, investigate the discrepancy. If the proposed explanation depends on an unvalidated assumption, acknowledge that uncertainty rather than presenting it as fact. And remember that no amount of technical sophistication removes our responsibility to listen, observe, follow up and exercise judgement.
I would be deeply concerned if HTMA practice evolved into little more than sending hair samples to a laboratory, feeding the results into an AI platform and forwarding the generated report to the patient. That may be an efficient information service, but it is not the kind of personalised healthcare I have spent my working life advocating.
We Must Retain the Human in the Equation
As I approach another birthday after more than half a century working in health, I find myself increasingly convinced that the fundamentals of good healthcare have not changed nearly as much as the technology surrounding them. People still need to be listened to. They need to be understood within the circumstances of their lives. They need advice that is proportionate, practical and open to revision as new information emerges.
Artificial intelligence may help us accomplish these things more effectively, but it can also tempt us to take shortcuts. We must resist the assumption that faster answers are necessarily better answers, or that a longer and more sophisticated report represents deeper understanding. I want a future in which AI gives health professionals more time to spend with people, rather than providing an excuse to remove the professional relationship altogether.
Behind every HTMA report is a person with a history, a family, concerns, ambitions and hopes for a healthier future. That person is far more complicated than any collection of mineral measurements. Let AI help us examine the evidence. Let it challenge our assumptions and improve our efficiency. But let us never confuse the machine's output with the entirety of clinical judgement.
The test provides clues. The person provides the context. And we must retain the human in the equation.
Gary Moller, DipPhEd, PGDipRehab, PGDipSportMed
Health, rehabilitation, sports medicine and nutrition practitioner since 1972.
This article is for education and discussion. HTMA has important limitations, and some proposed interpretations of hair mineral patterns lack established clinical validation. AI-generated reports should not be used alone to diagnose illness or determine treatment. Consult an appropriately qualified healthcare professional about symptoms, medications and significant changes to treatment.
Comments