The com­mon per­cep­tion is that AI can like­ly replace mod­ern med­i­cine, but not tra­di­tion­al med­i­cine. This is often based on the idea that mod­ern med­i­cine is stan­dard­ized and data-dri­ven, mak­ing it suit­able for AI, while tra­di­tion­al med­i­cine is per­son­al­ized and relies on sub­jec­tive assess­ments, mak­ing it dif­fi­cult for AI to han­dle. How­ev­er, this is a mis­con­cep­tion about tra­di­tion­al med­i­cine.

Tra­di­tion­al Chi­nese Med­i­cine (TCM), for exam­ple, is not based on ran­dom diag­noses. It also relies on diag­nos­tic indi­ca­tors and stan­dards to pre­scribe treat­ments. The fun­da­men­tal prin­ci­ple is the same as mod­ern med­i­cine. Do not mys­ti­fy TCM. TCM has its own diag­nos­tic indi­ca­tors, and based on the vari­a­tions in the com­bi­na­tion of these indi­ca­tors, dif­fer­ent con­di­tions are iden­ti­fied, and cor­re­spond­ing treat­ments and med­ica­tion com­bi­na­tions are pre­scribed.

Indi­­ca­­tor-con­di­­tion-treat­­ment is essen­tial­ly the same as mod­ern med­i­cine, and AI can def­i­nite­ly han­dle TCM. TCM should not be mys­ti­fied.

The chal­lenge for mod­ern peo­ple, espe­cial­ly those with edu­ca­tion, in accept­ing TCM lies in its mys­ti­fi­ca­tion by some. Stripped of its pre-mod­­ern ter­mi­nol­o­gy, TCM is essen­tial­ly a func­tion that maps symp­toms to treat­ments. It has stan­dards. What is the dif­fer­ence from mod­ern med­i­cine? For instance, pulse diag­no­sis in TCM can be stan­dard­ized. It is not dif­fi­cult to col­lect pulse wave pat­terns using sen­sors and then clas­si­fy them.

It essen­tial­ly assess­es whether a per­son is weak, their over­all con­di­tion, and whether their heart­beat is irreg­u­lar, and then decides whether to use spe­cif­ic med­ica­tions in appro­pri­ate amounts. It feels sim­i­lar to an elec­tro­car­dio­gram (ECG). Is ECG very use­ful for diag­no­sis every time you go to the hos­pi­tal? Isn't it a stan­dard pro­ce­dure, done every time because it can rule out some major risks.

TCM pulse diag­no­sis actu­al­ly looks at three indi­ca­tors: the reg­u­lar­i­ty of the pulse, whether it is fast or slow, and whether the pulse is strong. These three indi­ca­tors are com­bined into dif­fer­ent pulse pat­terns. Adding height and weight as addi­tion­al dimen­sions is not dif­fi­cult for a com­put­er. Many pulse-tak­ing robots already exist, but they are not yet wide­ly used in hos­pi­tals like ECG machines.

What is the dif­fer­ence between an AI robot and an ECG?

Essen­tial­ly, an ECG looks sci­en­tif­ic and mod­ern, mak­ing it feel worth­while to pay for. Pulse-tak­ing looks old-fash­ioned and unre­li­able, mak­ing it feel like a scam. This is an objec­tive cog­ni­tive bias, an old mind­set. Again, TCM also has sci­en­tif­ic indi­ca­tors, do not mys­ti­fy it. AI can let peo­ple know that TCM treat­ment has objec­tive stan­dards and its own log­ic, which is a good thing. So, can AI real­ly replace doc­tors? When AI + robots can per­form neu­ro­surgery, car­diac surgery, eye surgery, and micro­surgery, then it can replace 90% of doc­tors, which is the best-case sce­nario. The remain­ing 10% requires con­tin­u­ous inno­va­tion that can­not be found in any data­base. Anoth­er big­ger issue is that doc­tors, like accoun­tants, need to be respon­si­ble. AI can­not take respon­si­bil­i­ty, which means it ulti­mate­ly can­not replace doc­tors.