Why Did Theranos Collapse? The Limits of Fingertip Blood Sampling, Predicting Survival via Facial Scanning, and the Importance of Measuring Health Holistically
In this article
Seeing Everything in a Drop of Blood: The Theranos Promise and the Real Complexity of Health
Theranos became a 9-billion-dollar company in 2014 by promising to run more than 200 tests on a few drops of blood taken from a fingertip; its founder, Elizabeth Holmes, was later convicted of defrauding investors. The documentary "You Can See Everything", released in October 2026, has brought the story back into the spotlight. The lesson Theranos leaves behind is not only a tale of fraud: the very idea that a single device, a single blood test or a single face scan can measure health "completely" is misleading. Health is a whole, with biological, psychological and social dimensions, heredity and the microbiome in our gut. In this article we explain what can actually be done with small blood samples, why it is so difficult, and why we need to measure health holistically. We have discussed the Check The Life philosophy, the lessons we draw from the blue zones and Harvard's lifestyle medicine approach in separate articles.
- What Theranos promised: More than 200 tests from a few drops of finger-prick blood. In reality, the company's own device could run only about 12 tests; the rest were performed on commercial machines.
- Why it is hard: Drops taken from a fingertip can give noticeably different results from one another, and each test needs its own preparation, its own tube and its own chemistry. Running many accurate tests on a small sample is still a subject of research today.
- There is real progress: There are devices that give results in 12 minutes in the clinic and research methods that measure thousands of molecules in 10 µL of blood; but the clinical benefit of each must be proven separately.
- The real lesson: Neither a drop of blood, nor a smartwatch, nor a face scan shows the whole of health. Health requires blood, fitness, heart, mood, relationships, sleep, nutrition, heredity and the microbiome to be assessed together.
Why does this matter?
Every day a new device, app or test appears with the promise "learn about your health from a single measurement": hundreds of values from finger-prick blood at home, a "biological age" from a device on your wrist, dozens of diseases from a single blood sample. Some of these technologies are genuinely valuable; but the promise of "seeing everything" can, as with Theranos, be marketed long before the evidence exists. People may also reduce a complex health picture to a single number and end up either falsely reassured or needlessly worried (how the public is misled by seemingly sound research).
Health is not like other areas of information technology. The full state of a phone or computer can be measured with sensors; a human being, however, is a system that changes together with its biology, mood, relationships, genes and the trillions of microorganisms living within it. Preventing and managing chronic diseases, cancer above all, also requires these dimensions to be addressed together (a holistic approach to the cancer patient).
Can a single blood test or device fully measure health?
No. A blood test, a smartwatch or a face scan shows only one window on health. Blood values tell us about metabolism, but they do not show fitness, mental health, social relationships, sleep patterns, hereditary risk or the gut microbiome. Understanding health requires these dimensions to be measured together, interpreted together and followed over time.
What did Theranos promise, and what happened?
Theranos claimed it would run hundreds of tests, from cholesterol to cancer markers, quickly and cheaply on a few drops of blood taken from a fingertip. For people afraid of having blood drawn from a vein, and for those who need frequent testing, this was an extremely attractive promise. However, according to the 2018 complaint by the US Securities and Exchange Commission (SEC), while the company offered patients more than 200 tests, its own device could run only about 12 of them. The rest were performed on commercial laboratory analysers, some of them machines modified to work with finger-prick blood.
Theranos: the promise and the reality
Number of tests. The grey bar shows the tests offered; the coloured bar shows those the company's own device could run.
According to the 2018 complaint by the US Securities and Exchange Commission (SEC), the remaining tests were run on commercial analysers, some of them modified to work with finger-prick blood.
In the documentary, Holmes lays out the parts of the Theranos device on a table and argues that the machine "can do any test"; but as the questions go deeper, it emerges that this meant it could do so "once the necessary chemistry has been developed". What researchers working in diagnostics most wanted to see in those years was published data to support these claims; in the words of Meaghan Bond of Rice University, "that data never came".
Why is getting 200 results from one drop so difficult?
1. Every drop is different
In their study published in the American Journal of Clinical Pathology in 2015, Bond and Richards-Kortum of Rice University collected individual drops from fingertips and compared haemoglobin, white blood cell and platelet values with venous blood. All measurements fluctuated more in finger-prick blood than in blood drawn from a vein. In one volunteer, the haemoglobin difference between two consecutive drops reached about 2 g/dL. Such a difference matters when deciding on a blood transfusion; it may matter less in situations such as screening for iron deficiency (iron deficiency and ferritin).
Figure 1 · Finger-prick blood: every drop is different
Drop-to-drop variability and the factors that increase it
2. Each test needs its own preparation
As Brian Cunningham of the University of Illinois explains, blood often goes through a preparation step before it is tested. For some tests the fluid surrounding the cells (plasma) is separated by centrifugation; for others the cells are broken open with chemicals; and measuring DNA or RNA requires an entirely different chemistry from measuring proteins. Even the substance added to the tube to prevent clotting affects the result: for example, a biochemistry analyser used in the clinic requires lithium heparin, because other substances such as EDTA interfere with its measurements. Running many tests on a single small sample means reconciling these different needs within the same drop.
Figure 2 · One drop, incompatible preparations
The different needs of plasma, whole blood, intracellular and genetic tests
What can actually be done today?
The collapse of Theranos did not stop legitimate research on small blood samples. But every advance consists of steps that are far more limited than what was promised, and each must be proven on its own:
| Technology | What can it do? | Its limitation |
|---|---|---|
| In-clinic biochemistry analyser (Piccolo Xpress) | 14 values, including glucose and creatinine, from a 100 µL sample in about 12 minutes | In practice at least 1.5 ml of blood is drawn from the patient, 15 times what the device uses |
| Automated benchtop testing device (Truvian) | FDA clearance in November 2025 and January 2026 for kidney tests and blood counts | Clearance covers venous blood only; finger-prick blood needs separate evidence |
| Pocket-sized reader (VPodDuo) | Connects to a phone to measure genetic fragments linked to infection and cancer | So far tested only with samples prepared in the laboratory, not with patient blood |
| Microsample multi-omics (Stanford, Snyder) | Measuring about 7,000 molecules in 10 µL of blood repeatedly over the course of a day | Most are not routine clinical tests; it does not give basic values such as LDL and HDL cholesterol; largely a research tool |
| Small-volume tubes in intensive care (STRATUS) | Reducing unnecessary blood loss; randomised trial in 25 intensive care units | A small but not statistically significant reduction in the need for transfusion |
The information is compiled from the researchers' views and studies reported in the health news article "You Can't See Everything: What Theranos' Legacy Teaches Us".
Michael Snyder of Stanford describes his method as "a Theranos that works"; at the same time, he stresses that it is a research project and that more studies are needed to determine which measurements carry clinical value. Developing the method took 7 years. AI-supported approaches that aim to detect many diseases from a single blood sample are following a similar path (multiple diseases from a single blood sample, one blood sample, 98 diseases, the Galleri multi-cancer blood test). Each is promising; but the same question applies to each of them: does the value being measured actually change a patient's diagnosis, treatment or life?
Why can health not be captured in a single measurement?
The biopsychosocial model, used in medicine since 1977, holds that illness is not merely a biological event but is shaped together with mood, relationships and living conditions. Today we add two more dimensions to this picture: heredity and the microorganisms living in our gut (the microbiome). For example, low fitness is as strong an indicator of the risk of loss of life as smoking and high blood pressure; weak social ties are also associated with a risk of similar size. None of these shows up in a blood tube (exercise is a cancer medicine, emotional health and cancer risk).
Figure 3 · Health does not fit in a single drop
A blood test is only one of health's ten windows
- Biological dimension: Blood and biochemistry, blood pressure, heart rhythm, exercise capacity, muscle strength, body composition and visceral fat. Unlike chronological age, "biological age" is also part of this dimension (chronological and biological age, heart rate and longevity).
- Psychological dimension: Stress, burnout, anxiety and depression; these affect both quality of life and the ability to sustain healthy behaviours (do stress and anxiety cause cancer?).
- Social dimension: Family, friendship, belonging and support networks. Strong social ties are one of the common denominators of long life in the blue zones.
- Hereditary dimension: Family history and, where needed, genetic counselling; in some cancers this can change when and how screening is carried out (hereditary cancers).
- Host–microbiome dimension: Gut bacteria are linked to nutrition, immunity and drug response; they cannot be assessed with a blood test alone (our gut microbiota as a health compass).
- Behavioural dimension: Movement, sleep, nutrition, smoking and alcohol; measuring them only becomes meaningful when it turns into change (a sedentary lifestyle and ageing).
Can a face scan predict survival?
FaceAge, an artificial intelligence model developed by researchers at Mass General Brigham, estimates a person's "biological age" from a routine face photograph. In the study published in Lancet Digital Health in 2025, the face age of cancer patients was on average about 5 years higher than their chronological age, and a face that looked "older" was associated with shorter survival. In patients receiving palliative radiotherapy, doctors' estimates of 6-month survival improved markedly when FaceAge information was added. A follow-up study published in Nature Communications in 2026 showed that the pace of facial ageing between two photographs is also associated with survival.
What did a face photograph add to the doctor's estimate?
Accuracy in predicting 6-month survival in patients receiving palliative radiotherapy (AUC; 0.5 is a coin toss, 1 is perfect). The grey bar shows the doctors' estimate alone.
FaceAge, Lancet Digital Health 2025. The biological age estimated from the face photograph was assessed as information added to the doctor's estimate; it was not designed to guide diagnosis or treatment decisions on its own.
Figure 4 · Biological age from a face photograph
A valuable signal, but not the whole picture
These studies show that artificial intelligence can extract meaningful signals from unexpected data (predicting immunotherapy response with AI). But a face photograph, just like a drop of blood, is a single window: it does not tell us why a person is ageing faster, or whether this stems from poor fitness, lack of sleep, chronic stress or an illness. The difference between predicting and changing also explains why a holistic assessment is needed. This is also why the rules on how AI studies should be reported are becoming increasingly strict (JAMA's AI publication rules).
Check The Life: measuring, understanding and managing together
Check The Life starts from the exact opposite of the Theranos promise: rather than seeing everything with a single device, it measures the different dimensions of health through the eyes of different specialties and brings the findings together in a single report. The programme brings internal medicine, cardiology, psychology, nutrition and physiotherapy specialists to the same table. Alongside blood and biochemistry values, it assesses exercise testing, grip strength and functional tests, body composition and visceral fat, blood pressure and heart rhythm, sleep patterns, stress and burnout, social relationships, hereditary risk based on family history, and gut microbiome analysis.
Figure 5 · The Check The Life cycle
Holistic measurement, a shared report, 12 months of management together
The programme's motto, "Let's measure together, let's manage together", sums up this approach. Measurement is a starting point; the real value lies in turning the findings into sustainable changes in the person's life. This was also one of Theranos's greatest mistakes: it made measuring an end in itself, yet never managed to show how accurate its measurements were or what they would be useful for.
Figure 6 · Five questions for assessing a health technology claim
When faced with "one drop", "one device" or "one scan" promises
In Turkey, too, home finger-prick tests, devices that measure "biological age" and comprehensive "check-up" packages are spreading rapidly. Some of these tests are based on validated laboratory methods, while the accuracy and clinical benefit of others are uncertain. The fact that a test is registered as a medical device in Turkey does not mean its result is sufficient for a decision in every situation. Rather than interpreting results on their own, it is important to assess them with a doctor, together with the person's history, examination and other findings. Team-based, long-term follow-up programmes such as Check The Life are generally not covered by SGK (Turkey's Social Security Institution).
Advice for our readers
- When faced with promises of "hundreds of tests from a single drop of blood" or "complete health from a single device", ask for the published evidence.
- If a finger-prick test you have done at home gives an unexpected result, make decisions only after the test has been repeated on venous blood.
- Treat smartwatch and app data not as an alarm but as a trend you can discuss with your doctor.
- Do not neglect your fitness, sleep, mood and relationships just because your blood values are normal; these are measurable parts of your health too.
- If there is early-onset cancer or heart disease in your family, share this with your doctor; it may change your screening plan.
- Think of measurements not as a one-off snapshot but as a film to be followed over time; seeing change requires measuring again (evidence-based medicine).
Frequently asked questions
- Why did Theranos fail?
- Theranos promised to run more than 200 tests on a few drops of finger-prick blood, but its own device could run only about 12 tests. The product was marketed before scientific challenges, such as the drop-to-drop variability of finger-prick blood and the different preparation each test requires, had been solved, and data supporting its claims were never published.
- Is finger-prick blood as reliable as venous blood?
- It may be adequate for some tests, but finger-prick blood shows more variability from drop to drop. In one study, the haemoglobin difference between two consecutive drops reached about 2 g/dL. Tests that give a yes-or-no answer are less affected by this variability, whereas for tests that give a numerical value, venous blood is generally more reliable.
- Can a single blood test or smart device fully measure my health?
- No. Blood tests and smart devices provide valuable information but show only one dimension of health. Fitness, mental health, social relationships, sleep, hereditary risk and the gut microbiome should each be measured and interpreted together.
- What is survival prediction from a face photograph?
- An artificial intelligence model called FaceAge estimates biological age from a routine face photograph. In cancer patients, face age was shown to be about 5 years higher than chronological age, and an older-looking face was associated with shorter survival. This is information added to the doctor's assessment; it is not used on its own for diagnosis or treatment decisions.
Glossary
- Capillary (finger-prick) blood
- Blood from small vessels, obtained by pricking the fingertip.
- Venous blood
- Blood drawn with a needle from a vein in the arm; the standard sample for laboratory tests.
- Anticoagulant
- A substance added to a blood tube to stop the blood from clotting; heparin and EDTA are the most commonly used.
- Microsample
- A very small-volume blood sample, such as a few microlitres.
- Multi-omics
- Measuring thousands of molecules, such as genes, proteins and metabolites, together in the same sample.
- Biopsychosocial model
- A medical approach which holds that health and illness are determined jointly by biological, psychological and social factors.
- Microbiome
- The entire community of microorganisms, and their genes, living in the gut and other parts of the body.
- Biological age
- An estimated age showing how much the body has aged, independently of chronological age.
- AUC
- A measure of a test's or model's ability to predict correctly; 0.5 means a random guess and 1 means a perfect prediction.
DROZDOGAN Academy Commentary
The legacy of Theranos cuts both ways: on the one hand, it shows how a promise marketed without evidence can collapse; on the other, the goal of easier and more frequent testing with small samples remains worthwhile.
⚡ Why it matters
Although promises of "complete health from a single measurement" seem to offer people an easy way to manage their health, by reducing a complex picture to a single number they can create either false reassurance or needless anxiety.
✅ What we know today
There are devices that can run a limited number of tests on small samples during a consultation; thousands of molecules can be measured with microsamples; and even unexpected data such as a face photograph can give a signal about survival. The strong link between fitness, social ties and mental health and long life is also well established.
⚠️ Limitations
The variability of finger-prick blood, the differing preparation needs of different tests and the uncertain clinical meaning of many of the molecules measured remain unresolved. AI predictions do not explain causes. No single measurement shows the psychological, social, hereditary and microbiome dimensions of health together.
🩺 In practice
At Check The Life we measure health with five disciplines and across several dimensions, combine the findings in a single report and manage them together with the person for 12 months. For our cancer patients too, we regard assessing fitness, nutrition, mood and social support alongside treatment as part of their care.
❓ Open questions
Which of the thousands of molecules measured with microsamples will truly change clinical decisions? Can AI signals such as face photographs improve health outcomes when combined with a holistic assessment? How can the long-term benefit of holistic measurement and follow-up programmes be demonstrated in randomised trials?
Verdict: Health is not like other areas of information technology: no single device, single drop or single photograph shows everything. The right approach is to measure the different dimensions of health with the right methods, interpret them together and manage them together over time.
- This article was prepared in October 2026.
- Information about Theranos and the researchers' views are compiled from the health news article "You Can't See Everything: What Theranos' Legacy Teaches Us" and the studies it cites. The haemoglobin values in Figure 1 are examples.
- The FaceAge findings are based on reports about the Lancet Digital Health (2025) and Nature Communications (2026) publications; the ages in Figure 4 are examples.
- Conflict of interest statement: Check The Life is a programme founded by Prof. Dr. Mustafa Özdoğan, the owner of drozdogan.com. The programme's effect on long-term health outcomes has not yet been evaluated in a randomised trial.
- The illustrations are simplified. The Academy commentary is our own. This article is for general information only; it is not personal medical advice.
References
- U.S. Securities and Exchange Commission. Theranos, CEO Holmes, and former president Balwani charged with massive fraud. Press release, March 2018. sec.gov
- Bond MM, Richards-Kortum RR. Drop-to-drop variation in the cellular components of fingerprick blood: implications for point-of-care diagnostic development. Am J Clin Pathol 2015;144:885-894. doi:10.1309/AJCP1L7DKMPCHPEH
- Siegal DM, Belley-Côté EP, Lee SF, et al. Small-volume blood collection tubes to reduce transfusions in intensive care: the STRATUS randomized clinical trial. JAMA 2023;330:1872-1881. doi:10.1001/jama.2023.20820
- Shen X, Kellogg R, Panyard DJ, et al. Multi-omics microsampling for the profiling of lifestyle-associated changes in health. Nat Biomed Eng 2024;8:11-29. doi:10.1038/s41551-022-00999-8
- Bontempi D, Zalay O, Bitterman DS, et al. FaceAge, a deep learning system to estimate biological age from face photographs to improve prognostication: a model development and validation study. Lancet Digit Health 2025. PubMed
- Facial image analysis tracks biological aging, predicts cancer outcomes (Nature Communications 2026). HospiMedica, May 2026. hospimedica.com
- Engel GL. The need for a new medical model: a challenge for biomedicine. Science 1977;196:129-136. doi:10.1126/science.847460
- Nathan Fielder & Lance Oppenheim's Elizabeth Holmes doc "You Can See Everything". Motion Picture Association, September 2026. motionpictures.org
Was this article helpful?
Your feedback helps us improve our content.
